A Service of

Leibniz-Informationszentrum econstor Wirtschaft Leibniz Information Centre Make Your Publications Visible. zbw for Economics

National Bureau of Economic Research (NBER) (Ed.)

Periodical Part NBER Reporter Online, Volume 1995

NBER Reporter Online

Provided in Cooperation with: National Bureau of Economic Research (NBER), Cambridge, Mass.

Suggested Citation: National Bureau of Economic Research (NBER) (Ed.) (1995) : NBER Reporter Online, Volume 1995, NBER Reporter Online, National Bureau of Economic Research (NBER), Cambridge, MA

This Version is available at: http://hdl.handle.net/10419/62108

Standard-Nutzungsbedingungen: Terms of use:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Documents in EconStor may be saved and copied for your Zwecken und zum Privatgebrauch gespeichert und kopiert werden. personal and scholarly purposes.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle You are not to copy documents for public or commercial Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich purposes, to exhibit the documents publicly, to make them machen, vertreiben oder anderweitig nutzen. publicly available on the internet, or to distribute or otherwise use the documents in public. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, If the documents have been made available under an Open gelten abweichend von diesen Nutzungsbedingungen die in der dort Content Licence (especially Creative Commons Licences), you genannten Lizenz gewährten Nutzungsrechte. may exercise further usage rights as specified in the indicated licence. www.econstor.eu FAIL 1995

In This Issue Program Report: Program Report Monetary Economics 1

Research Summaries: The Role of Banks in the Transmission Monetary Economics of Monetary Policy 6 Behavioral Economics 9 Do Retirement Saving Programs Increase Saving? 13 N. Gregory Mankiw

NBER Proflles 18 Conferences 19 The NBER Program in Monetary Economics, established in 1991, en­ Bureau News 26 courages practical research in macroeconomics with an emphasis on is­ Bureau Books 37 sues relating to monetary policy. Its goal is to promote a greater under­ Current Worlting Papers 39 standing of the relationship between central-bank actions and the econo­ my in order to help the world's central bankers better meet the formida­ ble challenges they face. Toward these ends, program members pursue a range of individual and collaborative research studies. Program members and their guests meet twice a year to present and discuss this research. In addition, members meet for one week during the NBER's Summer In­ stitute to present recent research and collaborate on research in progress. Central bankers regularly are invited to attend these meetings and discuss their recent decisions and concerns. Over the past several years, program members have heard from Federal Reserve Governors Lawrence B. Lindsey and Janet L. Yellen; then President of the Federal Reserve Bank of Boston Richard Syron; then Governor of the Bank of Canada John Crow; former President of the Bundesbank Helmut SchleSinger; and the Bank of England's Chief Economist Mervyn A. King, among others. In addition to regular program meetings and the Summer Institute, the monetary economics program sponsors occasional conferences to focus research on specific topics. I ran the last such conference, which pro­ duced a recently published University of Chicago Press ·volume entitled Monetary policy. Christina D. Romer and David M. Romer currently are organizing the next conference, which will focus on policymaking in low­ inflation countries. Research by program members is diverse. It includes empirical and theoretical work on the effects of monetary policy and the study of alter­ native policies and institutional arrangements. What follows is a brief de­ Scription of some avenues of research that program members have been pursuing over the past several years.

NBER Reporter Fall 1995 The Effects of rter Monetary Policy ·NBERRep0 When the central bank changes • NATIONAL BUREAU OF ECONOMIC RESEARCH the quantity of money, what is the effect on the economy? Although this The National Bureau of Economic Research is a private, nonprofit research question is at the heart of mon­ organization founded in 1920 and devoted to objective quantitative analysis etary economics, the answer is nei­ of the American economy. Its officers and board of directors are: ther easy to obtain nor widely agreed upon. From a SCientific President and Chief Executive Officer-Marttn Feldstetn Director of Finance-Sam Parker point of view, the fundamental problem is that central banks do BOARD OF DIRECTORS not conduct controlled experi­ ments. If central banks randomized Chairman-Paul W. McCracken Vice Chairman-John H Btggs their actions, we could just observe Treasurer-Gerald A. Polansky what happened to the economy after these actions in order to see DIRECTORS AT lARGE their effects. But, in fact, central Peter Aldrich George C. Eads Robert T. Parry banks most often act in response Elizabeth E. Bailey Martin Feldstein Peter G. Peterson to actual, perceived, or anticipated John Herron Biggs George Hatsopoulos Richard N. Rosett Andrew Brimmer KarenN.Hom Bert Seidman events. Thus, researchers studying Carl F. Christ Lawrence R. Klein Kathleen P. Utgoff monetary policy confront the diffi­ Don R. Conlan Leo Melamed Donald S. Wasserman Kathleen B. Cooper Merton H. Miller Marina v. N. Whitman cult task of sorting out the effects Jean A. Crockett Michael H. Moskow John O. Wilson of central-bank actions from the causes of those actions. DIRECTORS BY UNIVERSITY APPOINTMENT Monetary economists have tried • George Akerlof, Calt/ornta, Berkeley Joel Mokyr, Northwestern to solve this problem in two ways. Jagdish W. Bhagwati, Columbta Andrew Postlewaite, Pennsylvanta One is to take a narrative approach William C. Brainard, Yale Nathan Rosenberg, Stanford Glen G. Cain, Wisconstn Harold T. Shapiro, Princeton to monetary history. This approach Franklin Fisher, MIT Craig Swan, Mtnnesota dates back at least to Milton Fried- Saul H. Hymans, Mtchtgan David B. Yoffie, Harvard Marjorie B. McElroy, Duke Arnold Zellner, Chtcago man and Anna]. Schwartz's classic treatise, A Monetary History of the DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS United States. This NBER-sponsored study examined in detail the causes Marcel Boyer, Canadtan Economics Assoctatton Mark Drabenstott, American Agricultural Economtcs Assoctatton and effects of major monetary William C. Dunkelberg, Nattonal Assoctatton of Bustness Economists changes over the previous century, Richard A. Easterlin, Economic History Assoctatton Gail Fosler, The Conference Board with an emphasis on historical and A. Ronald Gallant, American Statistical Assoctatton institutional detail. More recently, Robert S. Hamada, American Ftnance Assoctatton Charles lave, American Economtc Assoctatton Romer and Romer have extended Rudolph A. Oswald, American Federatton ofLabor and Congress this narrative approach. They used ofIndustrtal Organtzattons Gerald A. Polansky, American Instttute of Certf/ied Public Accountants the minutes of Federal Open Mar- Josh S. Weston, Commttteefor Economic Development ket Committee meetings to pin- point dates at which the Fed Contributions to the National Bureau are tax deductible. Inquiries concerning changed policy toward a tougher contributions may be addressed to Martin Feldstein, President, NBER, 1050 Massachusetts Avenue, Cambridge, MA 02138-5398. stance on inflation. The Romer and The Reporter is issued for informational purposes and has not been reviewed Romer dates, as they have come to by the Board of Directors of the NBER. It is not copyrighted and can be freely be called, provide one concrete al- reproduced with appropriate attribution of source. Please provide the NBER's Public Information Department with copies of anything reproduced. beit controversial method for study- Preparation of the NBER Reporter is under the supervision of Donna Zerwitz. ing the effects of changes in mone- Requests for subscriptions, changes of address, and cancellations should be tary policy.! sent to Reporter, National Bureau of Economic Research, Inc., 1050 A second way of trying to disen­ Massachusetts Avenue, Cambridge, MA 02138-5398. Please include the cur­ rent mailing label. tangle the causes and effects of •

2. NBER Reporter Fall 1995 monetary policy is to take a more equilibrate the demand for money There remains considerable con­ econometric approach. That is, with the reduced supply. Higher troversy about how much this new rather than relying on a careful interest rates, in turn, reduce in­ lending view adds to our under­ reading of history, one can study vestment, as the higher cost of bor­ standing of the monetary transmis­ the effects of monetary policy by rowing discourages some potential sion mechanism. Various questions applying time-series analysis to investors. In addition, higher inter­ remain open to dispute. When bank macroeconomic data. To make this est rates depress asset values, such lending falls after a monetary con­ approach work, some identifying as the prices of equities, which re­ traction, as indeed it does, how assumption is necessary to sort out duces household wealth and con­ much of the fall is attributable to cause and effect. One might as­ sumer spending, Thus, when the reduced supply because of the fall in sume, for instance, that changes in central bank reduces the money reserves and how much is attribut­ short-term interest rates not explic­ supply and raises interest rates, it able to reduced demand because able by other macroeconomic vari­ reduces consumption and invest­ of the overall economic contrac­ ables reflect changes in the prefer­ ment and, thereby, the aggregate tion? When the banking system ences of the central bank toward demand for gooqs and services. In finds itself short of reserves, to inflation. Arguably, this change the long run this. reduced demand what extent can it make loans in represents an essentially random leads to lower inflation, but in the other ways, such as by selling gov­ change in policy. Once such an short run it also depresses the lev­ ernment securities or issuing non­ identifying assumption is made, els of production and employment reservable certificates of deposit? When firms find themselves unable one can use macroeconomic data Or so goes the conventional wis­ to obtain bank loans, to what to trace out the effects of monetary dom. Recently, a group of econo­ extent are they able to borrow in shocks over time. One drawback mists has suggested that this story of other ways, such as through of this econometric approach is the monetary transmission mecha­ finance companies or trade credit? that it relies on identifying assump­ nism is incomplete at best, for it tions that are usually open to dis- These and many related questions ignores the key role of bank lending. have provided a fertile area of (I(): pute. On. the other hand, com­ According to this new view, called ~, pared to the narrative approach, research for members of the mone­ the lending view, when the central tary economics program. 3 the econometric approach asks for bank conducts open-market opera­ less subjective judgment on the tions to reduce" the supply of money, part of the researcher, which in­ it drains reserves from the banking Short-Run Price creases the verifiability of the re­ system, which in turn forces banks to Adjustment sults and reduces the possibility of reduce lending. 1his reduced capaci­ inadvertent bias. 2 According to conventional mon­ ty of the banking system to make etary theory, changes in the quanti­ loans provides another reason, in ty of money have important effects Banks and the addition to higher interest rates, that on production and employment Monetary Transmission investment falls after a monetary over a period of several years, but Mechanism contraction. According to this lend­ these effects dissipate over time. ing view, a monetary contraction has This distinction between the short­ In recent years, much effort has an especially strong impact on those run and long-run effects of mone­ been devoted to reexamining the investors who are bank-dependent tary policy rests on the assumption monetary transmission mecha­ Small finns, for instance, have limited that wages and prices respond dif­ nism-the way in which central­ ability to finance investment through ferently over different time hori­ bank actions affect the aggregate other means, such as the issuance of zons. Over short periods of time, demand for goods and services. stocks or bonds, and have . limited wages and prices are relatively According to conventional theory, cash flow to finance investment sticky; this inability to respond monetary policy affects aggregate through retained earnings. When the fully to changes in .the quantity of demand primarily through interest central bank reduces bank reserves money induces changes in real rates. When the central bank re­ and causes a contraction in bank economic activity. Over long peri­ duces the quantity of money lending, these bank-dependent firms ods of time, wages and prices can .' • through open-market operations, have little choice but to reduce respond to changing circum­ ~'-. for instance, interest rates rise to investment spending. stances, so changes in the quantity

NBER Reporter Fall 1995 3. of money have little effect on pro­ rules as guidelines when choosing policy is by studying the poliCies of duction and employment. how to exercise those powers. the past. Thus, monetary econom­ . I Thus, understanding the dynam­ There are a large number of ics is tied inextricably to monetary history. As noted earlier, the NBER • ic effects of monetary policy re­ possible policy rules from which a quires an understanding of how central bank might choose. A poli­ has a strong tradition in the narra­ wages and prices adjust over time. cy rule long advocated by some tive approach, including the classic Much research in the monetary economists is a target path for a work of Friedman and Schwartz economics program is aimed at this monetary aggregate, such as M2 or and the more recent work of Ro­ topic. Some of this work tries to the monetary base. Another possi­ mer and Romer. The monetary build theories that explain how bility is an adjustable target path economics program also includes Hrms choose to adjust prices. Other for a monetary aggregate, where many other researchers examining work examines data on prices, the adjustments occur according to historical monetary policies, from sometimes at the economywide some formula as the central bank the most recent recession to the level and sometimes at the level of collects new information about the policies and institutions of cen­ 7 individual Hrms, in order to evalu­ velocity of money. Another possi­ turies past. ate these theories and to shed light ble rule is a target path for nominal on actual price adjustment. 4 income, perhaps implemented us­ Overall Philosophy ing the consensus of private-sector forecasts of nominal income. Re­ The NBER monetary economics Policy Design, Rules, search can shed light on the pros and program is eclectic in the best Institutions, and cons of these and other rules, for sense of the word. It embraces History instance by simulating how history high-quality research on monetary policy and related issues, both the­ Economists who study monetary might have been different if one of these rules had been in effect. 5 oretical and applied, from a variety economics are motivated by more of perspectives. Members of the than the intrinsic intellectual inter­ Closely related to the study of policy rules is the study of institu­ program often disagree with one est of the Held. To a large extent, another. Indeed, it is the disagree­ tional design. It is now widely ac­ their goal is to provide a greater ments that lead to debate, research, understanding of the economy in cepted that central banks with greater independence from the and, ultimately, progress in eco­ • order to improve the conduct of nomic science. monetary policy. Thus, much work government tend to produce less in the NBER monetary economics inflation than central banks with program is aimed at evaluating al­ close governmental ties. This fact ternative approaches to policymak­ raises many questions. Are inde­ ISee M. Friedman andA.J Schwartz, A Monetary History of the United States, ing at the world's central banks. pendent central banks less infla­ 1867-1960, Princeton: Princeton Uni­ Some of this work considers the tionary because they care less versity Press, 1963;· C. D. Romer and D. effects of various policy rules. A about short-run fluctuations in pro­ H. Romer, "Does Monetary Policy monetary policy rule is a type of duction and employment? Are their Matter? A New Test in the Spirit of policy choices less affected by up­ Friedman and Schwartz, " in Macroeco­ contingency plan for the central nomics Annual, Volume 4, o. J. Blan­ bank: it speciHes how the central coming elections and other political chard and S. Fischer, eds. Cambridge: bank will respond to changing eco­ considerations? Are they more like­ MIT Press, 1989; and M. D. Shapiro, nomic circumstances. Policy rules ly to follow policy rules? Do the "Federal Reseroe Policy: Cause and Ef- are of interest for two reasons. lower rates of inflation of indepen­ fect, " NBER Reprint No. 1925, Decem­ dent central banks affect long-run ber 1994, and in Monetary Policy, N. First, some economists have advo­ G. Mankiw, ed. Chicago: University of cated reduced reliance on discre­ growth rates? Which of the many Chicago Press, 1994. institutional features of indepen­ tion and greater reliance on rules in 2 For some recent econometric studies of order to reduce the political pres­ dent central banks are most impor­ the effects of monetary policy, see R. G. sure and inflationary bias often tant? These and the many related King and M. W. Watson, "Testing Long­ considered inherent in discre­ questions have been the subject of Run Neutrality, " NBER Working Paper much recent research.6 No. 4156, September 1992; T. Konishi, tionary policy. Second, even if a V. A. Ramey, and C. W. J Granger, central bank maintains its discre­ The most direct way in which "Stochastic Trends and Short-Run Rela­ tionary powers, it can use policy research sheds light on monetary tionships Between Financial Variables • 4. NBER Reporter Fall 1995 "t-, and Real Activity, " NBER Working Pa- Black Box: The Credit Channel of Mon­ of Changing Financial Markets, " NBER - per No. 4275, February 1993; K. K. etary Policy Transmission," NBER Reprint No. 1889, june 1994; B. T. Mc­ Lewis, ''lire Foreign Exchange Interven­ Working Paper No. 5146, May 1995. Callum, "Monetary Policy Rules and Fi­ tion and Monetary Policy Related and 4For some recent work on price adjust­ nanci,al Stability, " NBER Working Paper Does It Really Matter?" NBER Working ment, see A. S. Blinder, "On Sticky No. 4629, january 1994; M. D. Bordo Paper No. 4377, june 1993, and Jour­ Prices: Academic Theories Meet the Real and A. J. Schwartz, "The Specie Stan­ nal of Business (April 1995); S. Fischer, World," in Monetary Policy, op. cit.; L. dard as a Contingent Rule: Some Evi­ "The Role of Macroeconomic Factors in M. Ball and N. G. Mankiw, "Relative dence for Core and Peripheral Coun­ Growth," NBER Working Paper No. Price Changes as Aggregate Supply tries, 1880-1990," NBER Working Pa­ 4565, December 1993; L. J. Christiano, Shocks, " NBER Reprint No. 1961, March per No. 4860, September 1994; and S. M. S. Eichenbaum, and C. L. Evans, 1995; M. F. Bryan and S. G. Cecchetti, G. Cecchetti, "Inflation Indicators and "The Effects of Monetary Policy Shocks: "Measuring Core Inflation, " NBER Inflation Policy, " NBER Working Paper Some Evidence from the Flow of Working Paper No. 4303, March 1993, No. 5161, june 1995- Funds, " NBER Working Paper No. and in Monetary Policy, op. cit.; L. M. 6 For some of the recent work on central 4699, April 1994; B. T. McCallum, Ball, "What Determines the Sacrifice bank independence, see B. Eschweiler "Monetary Policy and the Term Struc­ Ratio?" NBER Working Paper No. 4306, and M. 'D. Bordo, "Rules, Discretion, ture of Interest Rates, " NBER Working March 1993, and in Monetary Policy, and Central Bank Independence: The Paper No. 4938, November 1994; B. S. op. cit.; L. M. Bal( and N. G. Mankiw, German Experience, 1880-1989," Bernanke and I. Mihov, "Measuring "A Sticky-Price Manifesto, " NBER NBER Working Paper No. 4547, Novem­ Monetary Policy, " NBER Working Paper Working Paper No~ 4677, March 1994, ber 1993, and in Varieties of Monetary No. 5145, june 1995; and J. H. Coch­ and Carnegie-Rochester Conference Reform: Lessons and Experience on the rane, "Identifying the Output Effects of Series on Public Policy, forthcoming; B. Road to Monetary Union, P. Siklos, ed. Monetary Policy, " NBER Working Paper T. McCallum, ''li Semiclassical Model of Boston: Kluwer Academic Publishers, Price Level Adjustment, " NBER Working No. 5154, june 1995. 1994; S. Fischer, "Modern Approaches 3 For an overview of the literature on the Paper No. 4706, April 1994; S. Basu, to Central Banking, " NBER Working /ending view, see A. K. Kashyap and J. "Intermediate Goods and Business Paper No. 5064, March 1995; and B. T. Cycles: Implications for Productivity C. Stein, "Monetary Policy and Bank McCallum, "Two Fallacies Concerning and Welfare, " N,BER Working Paper No. Lending," NBER Working Paper No. Central Bank Independence, " NBER 4817, August 1994; A. K. Kashyap, 431 7, April 1993, and in Monetary Pol- Working Paper No. 5075, March 1995. , icy, op. cit. For other recent work in this "Sticky Prices: New Evidence from 7For some of the recent work on mone­ • area, see A. K Kashyap, O. A. Lamont, Retail Catalogs, " NBER Working Paper tary history, see C. W. Calomiris, ( and J. C. Stein, "Credit Conditions and No. 4855, September 1994, and , JI "Greenback Resumption and Silver the Cyclical Behavior ofInventories: A Quarterly Journal of Economics 110, 1 Risk: The Economics and Politics of Case Study of the 1981-2 Recession," (1995); J. J. Rotemberg, "Prices, Output, Monetary Regime Change in the United NBER Working Paper No. 4211, Novem­ and Hours: An Empirical Analysis ber 1992, and Quarterly Journal of Based on a Sticky Price Model, " NBER States,1862-1900," NBER Working Pa­ Economics 109, 3 (1994); v. A. Ramey, Working Paper No. 4948, December per No. 4166, September 1992; M. D. "How Important Is the Credit Channel 1994; and M. S. Kimball, "The Bordo, "The Gold Standard, Bretton in the Transmission of Monetary Poli­ Quantitative A nalytics of the Basic Woods, and Other Monetary Regimes: cy?" NBER Working Paper No, 4285, Neomonetarist Model, " NBER Working An Historical Appraisal, " NBER Reprint March 1993, and Carnegie-Rochester Paper No. 5046, February 1995. No. 1818, October 1993, and in Feder­ Series on Public Policy (Fall 1993); J. 5For some work discussing monetary al Reserve Bank of St. Louis Review 75, A. Miron, C. D. Romer, and D. N. Wei/, policy rules, see M. Feldstein and J. H. 2 (March/April 1993); M. Feldstein, "Historical Perspectives on the Moneta­ Stock, "The Use of Monetary Aggregates "Monetary Policy and Inflation in the ry Transmission Mechanism, " NBER Re­ to Target Nominal GDp, " NBER Work­ 1980s: A Personal View, " NBER Work­ print No. 1931, january 1995, and in ing Paper No. 4304, March 1993, and ing Paper No. 4322, April 1993, and in Monetary Policy, op. cit.; C. D. Romer in Monetary Policy, op. cit.; R. E. Hall American Economic Policy in the and D. H. Romer, "Credit Channel or and N. G. Mankiw, "Nominal Income 1980s, M. Feldstein, ed. Chicago: Uni­ Credit Actions? An Interpretation of the Targeting, " NBER Reprint No. 1916, Oc­ versity of Chicago Press, 1993; M. D. Postwar Transmission Mechanism," tober 1994, and in Monetary Policy, op. Bordo, E. U. Choudhri, and A. ]. NBER Reprint No. 1880, june 1994; A. cit.; M. Woodford, "Nonstandard Indi­ Schwartz, "Could Stable Money Have K. Kashyap andJ. C. Stein, "The Impact cators of Monetary Policy: Can Their Averted the Great Contraction?" NBER of Monetary Policy on Bank Balance Usefulness Be judged from Forecasting Working Paper No. 4481, October Sheets, " NBER Working Paper No. 4821, Regressions?" in Monetary Policy, op. 1993; C. M. Betts, M. D. Bordo, and A. August 1994, and Carnegie-Rochester cit.; B. T. McCallum, "Specification and Redish, ''li Small Open Economy in De­ Conference on Public Policy, forthcom­ Analysis of a Monetary Policy Rule for pression: Lessons from Canada in the ing; R. G. Hubbard, "Is There a Credit japan," NBER Reprint No. 1854, March 1930s, " NBER Working Paper No. 4515, Channel for Monetary Policy?" NBER 1994, and Bank of Japan Monetary November 1993; M. D. Bordo, H. T. Working Paper No. 4977, December and Economic Studies 1, 2 (November Rockoff, and A. Redish, '~Comparison 1994, and Federal Reserve Bank of St 1993); B. M. Friedman, "The Role of of the United States and Canadian { ',. Louis Review (May/june 1995); ~nd B. judgment and Discretion in the Con­ Banking Systems in the Twentieth Cen­ \"" S. Bernanke and M. Gertler, "Inside the duct ofMonetary Policy: Consequences tury: Stability Versus Efficiency, " NBER

NBER ReporterFall 1995 5. Working Paper No. 4546, November Wo0d5 International Monetary System, " Bernanke, "Tbe Macroeconomics of the • 1993, andJournal of Economic History NBER Working Paper No. 4642, Febru­ Great Depression, " NBER Working Pa-' 54 (June 1994); C. W. Calomiris, "Is the ary 1994; C. D. Romer and D. H. Ro­ per No. 4814, August 1994; and D. R. Discount Window Necessary? A mer, "What Ends Recessions?" NBER Re­ Feenberg and J. A. Miron, "Improving Penn-Central Perspective," NBER print No. 1930, December 1994, and in the AcceSsibility of the NBER's Historical Working Paper No. 4573, December NBER Macroeconomics Annual, Vol­ Data, " NBER Working Paper No. 5186, 1993; M. D. Bordo, D. Simard, and E. ume 9, S. Fischer and J. J. Rotemberg, July 1995. N. White, "France and the Bretton eds. Cambridge: MIT Press, 1994; B. S.

Research Summaries

The Role of Banks in the Transmission of Monetary Policy

Ani! K Kashyap and and nonfmandal firms, so that our Aggregate Patterns Jeremy c. Stein research naturally brings together central issues in corporate finance In our first paper, which was co­ Over the last five years, we have and macroeconomics. authored with David Wilcox, we been working on the question of Our key empirical findings can looked at aggregate data. Perhaps how the effects of monetary policy be summarized briefly by the fol­ the simplest aggregate empirical -that is, open market operations lowing picture of monetary policy implication of the bank-centric • -are transmitted to the economy. transmission: when the Fed tight­ view of monetary transmission is . The principal theme of our work ens policy, aggregate lending by that banks' loans should be corre- . has been that, in order to fully un­ banks gradually slows down, and lated closely with measures of eco­ derstand the monetary mechanism, there is a surge in nonbank financ­ nomic activity. In fact, there is a one must move beyond traditional ing (for example, the use of com­ strong correlation between bank macro explanations that emphasize mercial paper). When this substitu­ loans and unemployment, GNP, households' relative preferences for tion of financing is taking place, and other key macroeconomic in­ "money" and other less liquid as­ aggregate investment is cut back dicators, as demonstrated by Ber­ 1 sets. In our view, the role of the (by more than would be predicted nanke and Blinder. But, in tenus commerdal banking sector is cen­ solely on the basis of rising interest of establishing support for a "bank tral to the transmission ofmonetary rates). Small firms that do not have lending channel," such correlations policy. More specifically, two key significant buffer stocks of cash are are inconclusive because they factors shape the way in which most likely to trim investment (par­ could arise even if the lending chan- monetary policy works: the extent ticularly investment in inventory) nel were not operative. For exam- to which banks rely on reservable around the periods of tight money. ple, it may be that the correlations deposit financing, and hence adjust Similarly, smaller banks seem more are driven by changes in the de­ their loan supply schedules in the prone than large banks to reduce mand for bank loans rather than wake of open market operations; their lending. Overall, the results . the supply of bank loans. That is, and the extent to which certain suggest that monetary policy may bank loans and inventories might borrowers in the economy are have important real consequences, move together because banks al­ "bank-dependent," and therefore but not because of standard interest ways stand willing to lend, and cannot offset these Fed-induced rate effects. firms finance desired changes in lev- shifts in bank loan supply easily. el of inventories with bank loans. These two factors in turn depend The main point of the first paper on the details of the finandng ar­ was a way to overcome this funda- •. rangements available to both banks mental difficulty in separating the i

6. NBER Reporter Fall 1995 t role of loan demand from loan between bank loans and commer­ In our second paper, written supply. Our insight was that move­ cial paper arises not because of with Owen Lamont,9 we took a mi­ ments in substitutes for bank fi­ loan supply effects, but rather be­ cro perspective and compared the nancing should contain information cause of heterogeneity in loan de­ differences in inventory investment about the demand for bank financ­ mand. They claim that large firms of publicly traded companies that ing. For example, if bank loans are that typically use commercial paper do and do not have bond ratings. falling while the issuance of com­ financing tend to increase all forms The nonrated companies are typi­ mercial paper is rising, then we in­ of borrowing, while smaller firms cally much smaller than the rated fer that the supply of bank loans that are mostly dependent on companies, and are likely to be has contracted.2 Thus, we exam­ banks receive less of all types of fi­ bank dependent. Because of the ined movements in the "mix" be­ nancing. In Kashyap, Stein, and myriad of evidence suggesting that tween bank loans and loan substi­ Wilcox7 we show that such hetero­ Federal Reserve policy was restric­ tutes following changes in the geneities at best can account only tive prior to 1982, we began the stance of monetary policy.3 Using partially for our mix results-very study with an examination of the both the federal funds rate and the similar results o1?tain when we re­ 1982 recession. We found that dur­ Romer and Rome.-4 policy proxy as strict our tests tq, a sample of only ing this episode, the inventory indicators of the stance of mone­ large firms. Nonetheless, the basic movements of the nonrated com­ tary policy, we found that when point that heterogeneity could ren­ panies were much more sensitive the Fed tightens, the issuance of der the interpretation of our results to their own cash holdings than were the inventory changes for rat­ commercial paper surges while more ambiguous is valid. Accord­ ed companies. (In fact, there was loans (slowly) decline-that is, the ingly, in our subsequent research no significant liquidity effect for change in the mix indicates that we have designed tests using micro­ the rated companies.) loan supply has shifted inward. 5 level data to help clarify the interpre­ In contrast, we found that dur­ Having established that the bank tation of the aggregate correlations. ing subsequent periods there was .. fmandng mix could be used to in- little relationship between cash • fer movements in banks' loan sup­ Monetary Policy and holdings and inventory movements ply, we were able to test whether for the nonrated companies. For loan supply affected firms' spend­ Inventories: A Micro Perspective instance, during 1985 and 1986, ing. In other words, we were able when monetary policy arguably to check whether the correlation Perhaps the most exciting aspect was particularly loose, the correla­ between lending and spending oc­ of our initial aggregate findings is tion between inventory investment curred purely because of move­ that they may suggest a link be­ and cash holdings was negative ments in loan demand. Our tests tween monetary policy and inven­ and completely insignificant. boil down to checking whether the tory swings. It is well known that These findings further support proxy for shifts in loan supply has inventory de cumulations are large the view that financial factors be­ any additional explanatory power during recessions, and that mone­ yond those captured by open mar­ for investment once other funda­ tary policy is typically tight prior to ket interest rates play an important mental factors such as the cost of recessions. However, the simple role in shaping the effects of mone­ capital are taken into account. We story that tight money and high tary policy on inventory behavior. found that the mix does seem to carrying costs lead to inventory Related work by Gertler and Gil­ have independent predictive power runoffs is undermined by another christlO shows that the same type for investment, particularly for in­ well-known finding: the difficulty of results are found when one ventory investment. in documenting interest rate effects compares the inv~ntory investment While these results based on ag­ on inventories. Our initial results of an aggregated set of large and gregate data are fairly supportive of with Wilcox8 provide some support small firms: the inventory invest­ a lending channel, there are some for the view that monetary policy ment of small firms, which a priori important limitations that accompa­ and fmancial factors may be impor­ should be more susceptible to fi­ ny this type of time-series analysis. tant for inventory movements, even nancing problems, is more sensi­ For instance, Oliner and Rude- though standard interest rates in the tive to changes induced by mone­ • busch6 claim that the aggregate evi­ security market do not have much tary policy than is the investment of ,JI dence regarding the substitution predictive power for inventories. large firms.

NBER Reporter Fall 1995 7. Monetary Control via for their riskiness. In this case, giv­ than do ("large") banks that should e\ the Banking System en their imperfect information, in­ be able to raise external funds easi- ' dividuals may insist that all banks ly. Our most recent paper13 tests While it seems fairly clear that pay a large premium to attract this hypothesis using data at an in­ monetary policy derives some of its funds and safer banks may prefer termediate level of disaggregation. potency from some form of imper­ to make fewer loans rather than to Because of the complications of fection in the capital market at the pay the rate required to attract working with bank level data, we firm level, the precise connection funds. Alternatively, if the bank is opted as a first step to explore between monetary policy and bank able to fund itself completely with whether the behavior of a compos- loan supply is less clear. An oft­ insured deposits, then the adverse ite bank made up of all "large" asked skeptical question is why a selection problem is immaterial. banks differed from that of another Modigliani-Miller type of theorem Therefore, adverse selection prob­ composite bank made up of all does not hold for banking firms. lems can affect bank lending be­ "small" banks. Put differently, why should banks havior if the bank's ability to issue We found that small banks' respond to tight monetary policy insured deposits somehow is con­ lending was indeed more sensitive by cutting back on their loans, rath­ strained, and it has to turn to non­ to Fed-induced shocks to deposits er than by trying to continue lend­ insured sources of finance. than lending by large banks. We ing by raising more money in the One implication of the model is also found that the small banks' capital markets, or by disposing of that if the Federal Reserve can con­ holdings of securities were as sen­ other assets, such as securities?l1 trol the aggregate level of real in­ sitive to these shocks as the securi­ Part of this confusion is the result sured deposits available to banks, ties holdings of large banks. Over­ of the limited amount of theoretical then it will be able to influence all, the fact that small banks had to work on this topic; the solid micro­ loan: supply. Loosely speaking, reduce their asset holdings more economic arguments that justify the when the Fed drains reserves from across the board is consistent with conditions needed to permit mone­ the system, it forces banks to sub­ the notion that they face greater e tary policy to shift bank loan sup­ stitute away from insured deposit difficulties in adjusting to mone- , i ply had not been made when we financing and toward adverse-se­ tary-policy-induced contractions in started our work. Furthermore, em­ lection-prone forms of nondeposit deposit financing. Therefore, the pirical evidence that speaks direct­ finance. This in turn leads to an results fit with the adverse selec- ly to this phenomenon is hard to overall cutback in bank lending. In tion model of bank financing. come by. this sense, the model provides a We are now in the process of One of us recently developed a specific set of microfoundations for organizing a true bank-level data- fully articulated model that helps to the notion of a bank-centric view base that can be tapped to sharpen address some of the theoretical of monetary transmission. these tests. By studying micro data concerns.12 In the model, banks on banks' balance sheets, we can raise funds from individual inves­ Monetary Policy and identify with confidence banks that tors, and then in turn lend these should be most prone to cut lend- funds out to borrowers, who can Bank Balance Sheets ing in response to deposit shocks. be thought of as bank-dependent In addition to its macroeconom­ For instance, banks with low levels corporations. The typical bank's ic pre~ctions, the model empha­ of capital, banks that are chronic job is complicated by the fact that sizes that banks typically will face net borrowers in the federal funds individual investors are not as well the same sort of financing prob­ market, and small banks that have informed as bank management lems as other nonfinancial firms. no national name recognition all about the value of the bank's exist­ Accordingly, it suggests a range of can be examined to see if their ing assets. Depending on the type additional micro tests that we are lending is especially sensitive to of liability the bank issues to fi­ beginning to undertake. For in­ access to funds. The microdataset nance itself, this may create an ad­ stance, it suggests looking at also can be used to study more verse selection problem, because whether banks that should have general issues regarding bank be­ banks with lots of nonperforming trouble raiSing external finance­ havior, such as how banks' lending assets need to be charged a rela­ "small" banks-respond differently moved in response to the regional • tively high interest rate to account to a tightening of monetary policy economic shocks during the 1980s. , ~

8. NBER Reporter Fall 1995 We remain enthusiastic about move in opposite directions following a Movements in this spread help to fore­ our continued investigation of vari­ monetary contraction should not be ta­ cast investment even after· controlling ous aspects of bank behavior, be­ ken as an indication that firms cut off for the cost ofcapital. from banks are the ones who begin is­ 6S. D. Oliner and G. D. Rudebusch, '~ cause the implications of theories suing commercial paper. Instead, a Comment on Monetary Policy and Cred­ that place banks at the center of much more realistic interpretation is it Conditions: Evidence from the Com­ the monetary policy transmission that firms cut offfrom bank lending re­ position of External Finance, " Amer­ mechanism are quite different from ceive increased trade credit, and that ican Economic Review,forthcoming. those of standard theories that ig­ trade credit is supplied by firms with 7A. K. Kashyap, J C. Stein, and D. W. access to the commercial paper market. nore the role of banks. For exam­ Wilcox, "Monetary Policy and Credit See c. W. Calomiris, C. P. Himmelberg, Conditions: Evidence from the Composi­ ple, the presence of a lending and P. Wachtel, "Commercial Paper tion of External Finance: Reply," Amer­ channel implies that: and Corporate Finance: A Microeco­ ican Economic Revtew,forthcoming. • standard indicators, such as the nomic Perspective," NBER Working Pa­ 8A. K. Kashyap,]. C. Stein, and D. W. per No. 4848, September 1994, and quantity of money, do not cap­ Wilcox, "Monetary Policy and Credit Carnegie-Rochester Conference Series Conditions: Evidence from the Compo­ ture the effects of monetary pol­ on Public Policy 42 (1995), for some sition of External Finance, " American icy adequately; evidence showing -that this mechanism Economic ReView (March 1993), pp. • changes in the regulation of the may explain why b.,ank loans and com­ 78-98. mercial paper are substitutes at an ag­ 9A. K. Kashyap, O. Lamont, and]. C. banking sector can have impor­ gregate level. tant consequences for the trans­ Stein, "Credit Conditions and the Cycli­ 3We defined the mix variable to be the cal Behavior of Inventories, " Quarterly mission of monetary policy; and, ratio of bank loans to the sum of bank Journal of Economics 109 (1994), pp. • monetary policy may have distri­ loans plus commercial paper. 565-592. butional consequences that de­ 4C. D. Romer and D. H. Romer, "New 10M. Gertler and S. Gilchrist, "Monetary pend on how much finandng a Evidence on the Monetary Trans­ Policy, Business Cycles, and the Behav­ mission Mechanism, "Brookings Papers ior of Small Manufacturing Firms, " firm or sector receives from banks. on Economic ActiVity (1990), pp. 149- Quarterly Journal of Economics 109 Ultimately, we hope that our work 213. (1994), pp. 309-340. will allow us to learn enough about 5See E. M. Friedman and K. N. Kuttner, llSee A. K. Kashyap and J C. Stein, the role of banks in the transmis­ "Economic Activity and Short-Term "Monetary Policy and Bank Lending, " in Monetary Policy, N. G. Mankiw, ed. sion of monetary policy to gauge Credit Markets: An Analysis of Prices and Quantities, " Brookings Papers on Chicago: University of Chicago Press, the importance of these sorts of Economic Activity (1993), pp. 193-266, 1994, pp. 221-256, for further discus­ considerations. for a more comprehensive discussion of sion ofthese issues. this mechanism. Another way to use in­ 12See]. C. Stein, '~n Adverse-Selection formation about substitutes for bank Model of Bank Asset and Liability Man­ loans to resolve the identifu:ation prob­ agement with Implications for the IE. S. Bernanke and A. S. Blinder, "Cred­ lem is to study movements in relative Transmission of Monetary Policy, " it, Money, and Aggregate Demand," prices rather than relative quantities. NBER Working Paper No. 5217, August American Economic Review Papers Specifically, changes in loan supply 1995. and Proceedings 78 (1988), pp. 435- could be identifwd by checking whether 13See A. K. Kashyap and]. C. Stein, 439, and '7be Federal Funds Rate and the price of loans increases relative to "The Impact of Monetary Policy on the Channels of Monetary Transmis­ the price of an alternative such as com­ Bank Balance Sheets," NBER Working sion, " American Economic Review 82 mercial paper. We also found that when Paper No. 4821, August 1994, and Car­ (1992), pp. 901- 921. the Fed tightens, the prime rate rises rel­ negie-Rochester Conference Series on. 2A finding that these forms offinancing ative to the commercial paper rate. Public Policy 42 (1995), pp. 151-195.

Behavioral Economics

Richard H. Thaler are what distinguishes economics damental weakness in the method. from other social science disci­ The assumption of rationality is Neoclassical economics is built plines, such as psychology and so­ what permits economists to use on the assumption that the agents dology; as such, they are simulta­ their most powerful theoretical in the economy are self-interested neously the source of the power in tool: optimization. But, if agents and rational. These assumptions economic theorizing and the fun- are not rational, what then? My re-

NBER Reporter Fall 1995 9. search over the last 15 years has at- If economic agents make judg- any way for you to make money tempted to answer this question. ments and choices that differ sys- from my mistake, either by exploit- Of course, critiques of the as- tematically from those prescribed ing it or by educating me. Quasi- ." sumptions of economics are as old by the rational model, then we can rationality is neither fatal nor im- as the use of those assumptions. improve the models by incorporat- mediately self-defeating. For many economists, Milton Fried- ing these factors into our theories. How can we begin to model man provided the definitive re- My first paper in this domain4 de- "Homo Behavioral Economicus"? A sponse to such criticism in his fa- scribed several ways in which most small list of factors can go a long mous essay on positive econom- people fail to act like "Homo Eco- way toward better descriptive mod- ics. 1 There he argued that theories nomicus": for example, they fail to els. The following concepts have should not be evaluated on the ba- ignore sunk costs, they undervalue proven most useful so far: sis of the validity of their assump- opportunity costs relative to out-of- t. Overreaction. If people tions, but rather on the accuracy of pocket costs, and they have trou- make judgments using what Kahn- their predictions. An expert bil- ble exerting self-control. Some- eman and Tversky call the repre- liards player, he noted, may not times, agents know about their sentativeness heuristic, they judge know the laws of physics, but acts own biases. Thus, people who the likelihood of an event by how as if he knows such laws. I com- know they have self-control prob- similar the event is to the typical or pletely share Friedman's view that lerns will, like Odysseus, tie them- stereotype of that event. This heu- theories should be judged on the selves to the mast to prevent future ristic, like all those that are used basis of predictive power; in my transgressions. They join Christmas widely, is often accurate, but it research, I have used this criterion clubs (or used to before credit cards leads to systematic biases. For ex- to evaluate alternative models. eliminated the need to be liquid at ample, forecasts stemming from However, as I have tried to show Christmas time), go to fat farms (re- the representativeness heuristic vi- in my series of "Anomalies" articles sorts that, for a high fee, agree to olate Bayes's rule because recent in the Journal ofEconomic Perspec- starve their guests), and pay in ad- evidence is given too much weight tives, 2 the theory fails on precisely vance to join a health club, know- relative to prior odds. Forecasts these grounds. In such circum- ing that the sunk cost will help tend to be too extreme, relative to stances it makes sense to take a motivate them to go more often. the norm. • careful look at the assumptions. One objection to models with 2. Loss aversion. It is well ac- Perhaps most economic agents less than fully rational agents is the cepted that people tend to adapt to make decisions the way most of us claim that in markets, such agents current circumstances and to react play pool: badly. either will be eliminated or ren- to changes relative to the recent At some level, of course, the ra- dered irrelevant. Russell and I in- norm rather than to levels. Further- tionality assumption has to be vestigated this claim in a paper5 more, the sensitivity to losses is wrong: As noted by Herbert Si- that considers a world with two greater than the sensitivity to gains. mon, people are only boundedly kinds of agents: the fully rational Roughly speaking, the loss of $1 is rational. How does a boundedly ra- agents that populate standard eco- about twice as painful as the gain tional agent differ from a rational nomic models, and what we call of $1 is pleasurable. agent? If the differences are ran- "quasi-rational agents" who make 3. Mental accounting. Standard dom then the rational model still predictable, systematic errors. We accounting principles, and all eco- produces unbiased predictions of then tooked for the conditions nec- nomic theory, assumes that money behavior. However, as the work of essary for such a world to produce is fungible. Behavior should not Daniel Kahneman and Amos Tver- the same equilibriums that would depend on the label associated sky3 has shown, actual behavior obtain if all the agents were fully with any pot of money. However, differs from rational choice in sys- rational. We find that these condi- both individuals and organizations tematic ways. Their research pro- tions rarely are met, even in mar- violate this principle. This implies gram of discovering the heuristics kets that function very well, such that the rules people use to aggre- people use to make judgments, as financial markets. Often, if I in- gate transactions and wealth hold- and the biases inherent in those sist on choosing a less than opti- ings are important in understand- heuristics, provided my motivation mal choice for me (say a dominat- ing their behavior. I call these rules in exploring behavioral economics. ed alternative), there will not be "mental accounting.,,6 • 10. NBER Reporter Fall 1995 4. Fairness. All other things beha vioral explanation for the changes in earnings were too ex- t equal, people prefer to be treated "price/earnings" anomaly: namely, treme. One could improve the ana- fairly and like to treat other people that stocks with low p/e ratios sys- lysts' forecasts simply by subtract- fairly. This is not to say that people tematically outperform those with ing a constant (to correct for the do not care first and foremost high p/e's. His explanation was: optimism bias) and reducing the about their own household's wel- because investors use the represen- forecasted change by a third (to fare, but rather that concerns about tativeness heuristic, they would make the forecasted changes less fairness also matter and compete make excessively optimistic earn- extreme). with purely selfish motives for ings forecasts for firms growing Saving behavior is a domain in scarce resources. In another pa- rapidly and excessively pessimistic which behavioral factors have per,7 Kahneman, Knetsch, and I forecasts for those in trouble, that proven to be extremely important. explored lay perceptions of what is is, high and low p/e firms, respec- In the standard life-cycle model, fair. We found that the determi- tively. When these forecasts proved households compute their lifetime nants of such perceptions were ex- wrong in the predictable ways, wealth constraint, optimize their plained in tum by other behavioral prices would a,djust, causing high spending stream, and save accord- factors such as loss aversion. For p/e firms to ha.ve low returns and ingly. Components of wealth are example, people thought it was low p/e firms to have high returns. not distinguished since all money much less fair for a dealer to im- We thought that if this argument is assumed to be fungible. In the pose a $200 surcharge above list was correct we should be able to behavioral life-cycle model,12 the price for a scarce car model than to observe the same phenomenon for standard framework is modified to eliminate a $200 discount. Percep- firms chosen by past performance incorporate two important behav- tions of fairness also displayed rather than p/e. We therefore ioral factors: self-control and men- money illusion. For example, cut- ranked firms by 3-5-year past re- tal accounting. ting wages 5 percent in an econo- turns and selected the most ex- First, we recognize that most my with no inflation was judged to treme performers. We then tracked households other than the very ,t be much less fair than offering a 7 these extreme "winners" and "los- rich are constrained by self-control percent raise in an economy with ers" for the next 3-5 years. We considerations. Even if they could 12 percent inflation. found substantial mean reversion, and did solve for their optimal I have applied these and other as predicted by the behavioral spending path, they would find it behavioral factors in a variety of analysis. 10 difficult to stay on this path. Mental settings, always with the goal of De Bondt and I also looked for accounting comes into play be- trying to improve the explanatory evidence of the representativeness cause the location of wealth influ- power of economic analysis. One heuristic in professional analysts' ences spending choices, since domain that has taken a lot of my forecasts of earnings. 11 Recall that some mental accounts are more attention over the last ten years is when this heuristic is used, fore- "tempting" than others. For exam- financial markets. This field is an casts are too extreme. We re- pIe, cash on hand is more tempting attractive place to do behavioral gressed the actual change in earn- than money in a savings account, economics for two reasons. First, ings for a year on the consensus which in turn is more tempting the data are extraordinarily good. forecasted change in earnings than money in an IRA, pension Second, many economists have the (made in April for December 31 f18- plan, or home equity. "prior" (or did a decade ago) that cal year firms). If the forecasts were The difference between the behavioral factors are least likely to rational, the intercept in this regres- standard model and the behavioral surface in these, the most efficient sion would be zero and the slope model is particularly striking in of all markets. would be one. Instead we found analyzing programs designed to in- My first papers in this area were that the forecasts were biased in crease saving, such as lRAs and written with Werner De Bondt. In two ways. First, the intercept was 401(k)s. According to the standard the first one,8 we reported the suc- significantly negative, implying that theoretical analysis, these plans cessful prediction of a new anom- the forecasts were optimistic (con- have no effect on saving because aly in asset markets, namely long- sis tent with past research). Second, most contributors already had term mean reversion for individual the slope coefficient was ocly 0.65. some assets saved up that simply ,t stocks. David Dreman9 offered a This means that the forecasted could be shifted to the retirement

NBER ReporterFalll995 11. ! accounts. In this scenario, ther~ is no were roughly twice the reservation pension plans, endowments, and e\:1 I incentive to save at the margin, so the prices of the mug buyers. In other those saving for retirement) should programs should have no effect. words, we found "loss aversion" have horizons of much more than In a mental accounting frame- for mugs. Obviously, this result one year. work, however, saving is expected calls into question the Coasean As I hope these examples have to increase because the programs claim that in the absence of trans- made clear, I view behavioral eco- help put money into less tempting actions costs (known to be negligible nomics as an entirely constructive accounts. Consider a child who in this case) the initial assignment enterprise. The goal is simply to takes money from a leaky piggy- of property rights will not affect make economic models better at bank and puts it into a bank. Ac- the ultimate allocation of resources. explaining economic activity. In cording to the standard theory, as- In a paper with Benartzi,15 the this sense, behavioral economics is sets simply have been shifted, but concepts of loss aversion and men- simply economics with a higher R2. a behavioral analysis predicts that tal accounting are combined to of- spending will fall. The same is true fer an explanation for the well- for IRAs and other pension saving; known equity premium puzzle.16 1M. Friedman, "The Methodology of my reading of the evidence is that The puzzle refers to the fact that Positive Economics, " in Essays on Posi- this view is supported.13 over very long periods of time, tive EconOmiCs, Chicago: University of Chicago Press, 1953. Loss aversion is perhaps the equities have earned much higher 2These are reprinted in R. H. Thaler, rates of return than bonds or other most pervasive of all the behav- The Winner's Curse, Princeton: Prince- ioral factors uncovered to date. In fixed-income assets (roughly 6-7 ton University Press, 1992. a sample experiment Kahneman, percent real annual return versus 1 3D. Kahneman, P. Siovic, and A. Tver- Knetsch, and I ran,14 subjects trad- percent). This difference is too sky, Judgment Under Uncertainty: Heu- ed tokens, whose value to each large to be explained easily within ristics and Biases, New York: Cam- subject was private information. the standard framework. Our ex- bridge University Press, 1982; and D. Kahneman and A. Tversky, "Prospect Half of the subjects were then giv- planation is based on the following Theory: An Analysis of Decision Under en a token, and markets were intuition: suppose investors are Risk," Econometrica 47, 2, (1979), pp. "conducted" in which reservation loss averse, implying that losses are 363-391. .'i prices were elicited from both to- weighted more heavily than gainsP 4R. H. Thaler, "Toward a Positive Theo- ken owners and buyers. Both price Then, their willingness to hold ris- ry of Consumer ChOice, " Journal of and volume in these markets were ky assets will depend on the fre- Economic Behavior and OrganiZation 1 (1980) pp. 39-60. This paper, and most exactly as simple supply and de- quency with which they evaluate of the others cited here, are reprinted in mand analyses would predict, their holdings. If loss-averse inves- R. H. Thaler, Quasi-Rational Econom- showing that transactions costs tors compute the value of their ics, New York: Russell Sage Foundation, were negligible in this market. portfolios daily, they will hate 1991. Then, another set of markets equities, since on a daily basis 5r. Russell and R. H. Thaler, "The Rele- vance of Quasi-Rationality in Competi- was conducted, this time for coffee stocks decline nearly as often as tive Markets, " American Economic Re- mugs imprinted with the local uni- they rise. In contrast, investors with view 75 (December 1985), pp. 1071- versity insignia. Again, half the a horizon of say 20 years would 1082. subjects had a mug and half did find stocks very attractive since the 6R. H. Thaler, "Mental Accounting and not. The Coase theorem predicts risk of loss is tiny. Thus motivated, Consumer ChOice, " Marketing Science 4 (Summer 1985), pp. 199-214. that in this situation half°the mugs we estimate the frequency with which loss-averse investors would 7D. Kahneman, L. Knetsch, and R. H. on average should trade: mugs Thaler, "Fairness as a Constraint on should end up in the hands of the have to be evaluating their portfo- Profit Seeking: Entitlements in the Mar- subjects who value them mos t, lios in order to make the investors ket, " Americen Economic Review 76 only half of whom would have re- indifferent between stocks and (September 1986), pp. 728-741. ceived one in the random assign- bonds. We fmd this to be approxi- 8w. F. M. De Bondt and R. H. Thaler, ment of mugs at the beginning of mately once a year, a highly plausi- "Does the Stock Market Overreact?" Journal of Finance 40 (July 1985), pp. ble result. Therefore we dub our the experiment. Counter to this 793-805- explanation for the equity premium prediction, only about 15 percent 9D. Dreman, The New Contrarian In- of the mugs traded; the median puzzle "myopic loss aversion," vestment Strategy, New York: Random reservation prices of mug owners since most investors (for example, House, 1982. .1"

12. NBER Reporter Fall 1995 lOThis pattern has now been well doc­ American Economic Review (May rem," Journal of Political Economy 98 umented in numerous additional stud­ 1990), pp. 52-57. (December 1990),PP. 1325-1348. ies, such as W. F. M. De Bondt and R. 12R. H. Thaler and H. M. Shefrin, '~n 15S. Benartzi and R. H. Thaler, "Myopic H. Thaler, "Further Evidence on Inves­ Economic Theory of Self- Control, "J our­ Lass Aversion and the Equity Premium tor Overreaction and Stock Market Sea­ nal of Political Economy 89 (1981), pp. Puzzle," Quarterly Journal of Econom­ sonality, "Journal of Finance 42 (July 392-401; and H M. Shefrin and R. H ics 110 (February 1995),PP. 73-92. 1987), pp. 557-581; E. Fama and K. Thaler, "The Behavioral Life-Cycle Hy­ 16R. Mehra and E. C. Prescott, "The French, "The Crass-Section of Stock Re­ pothesis, "Economic Inquiry 26 (Octo­ Equity Premium Puzzle, "Journal of turns," Journal of Finance 46 (1992), ber 1988), pp. 609-643. Monetary Economics 15 (1985), pp. pp. 427-466; and]. Lakonishok, A. 13See, for example,]. Skinner "Individ­ 145-161. Shleifer, and R. W. Vishny, "Contrarian ual Retirement Accounts: A Review of 17Specifically, we assume investors act Investment, Extrapolation, and Risk, " the Evidence, "Tax Notes 54, 2 (Janua­ in accordance with cumulative pros­ Journal of Finance 49 (December ry 1992), pp. 201-212. pect theory. See A. Tversky and D. 1994), pp. 1541-1578. 14D. Kahneman, L. Knetsch, and R. H. Kahneman, '~dvances in Prospect The­ 11 W. F. M. De Bondt and R. H. Thaler, Thaler, "Experimental Tests of the En­ ory: Cumulative Representation of Un­ "Do Security Analysts Overreact?" dowment Effect and the Coase Theo- certainty," Journal of Risk and Uncer­ tainty 5 (1992), pp. 297-323.

Do Retirement Saving Programs Increase Saving?

David A. Wise ployee could contribute $2,000 per the return on the contributions ac­ year to an IRA account, and a nOn­ crues tax free; taxes are paid upon A large fraction of American working spouse could contribute withdrawal. But these plans are families reach retirement age with $250. The contribution was tax de­ available only to employees of virtually no personal financial as­ ductible. Annual contributions grew firms that offer such plans. Prior to sets. The median level of all per­ from about $5 billion in 1981 to 1987 the employee contribution sonal financial assets of families about $38 billion in 1986, approxi­ limit was $30,000, but the Tax Re­ with heads aged 55 to 64 was only mately 30 percent of total personal form Act of 1986 reduced the limit $8,300 in 1991; excluding Individu­ saving. Contributions declined pre­ to $7,000, and indexed this limit al Retirement Accounts (IRAs) and cipitously after the Tax Reform Act for inflation in subsequent years. 401(k) balances, the median was of 1986, even though the legislation The contribution limit is $9,235 for only $3,000. Almost 20 percent of limited the tax deductibility of COn­ the 1994 tax year. families had nO financial assets at tributions only for families who By 1991, contributions to all all. Thus, other than Social Security had annual incomes over $40,000 personal retirement saving plans benefits, employer-provided pen­ and who were covered by an em­ exceeded contributions to tradi­ sion benefits, and housing wealth tional employer-provided pension that is illiquid, the typical family ployer-provided pension plan. By 1990, less than $10 billion was COn­ plans. It seems evident that if it has very limited resources to use in were not for the 1986 tax legisla­ tributed to lRAs. Whereas over 15 meeting unforeseen expenses. In tion, personal retirement saving percent of tax ftlers made contribu­ addition to individual hardship, would have been much larger. tions in 1986, only 4 percent COn­ One consequence of the low U.S. Whether these programs increase tributed in 1990. saving rate is the prospect of limit­ net saving can be of critical impor­ ed future economic growth at the The other program, the 401(k) tance to future generations of older aggregate level. plan, grew continuously and al­ Americans and to the health of the Two saving programs introduced most unnoticed, with contributions economy in general. The issue re­ in the early 1980s were intended to increasing from virtually zero at the mains an important question of encourage individual saving. lRAs beginning of the decade to over economic debate. In a series of pa­ rapidly became a very popular $51 billion in 1991. In 1991, almost pers based on very different meth­ form of saving in the United States 25 percent of families contributed ods of analysis, Steven F. Venti and after they were made available to to a 401(k). Deposits in 401(k) ac­ I and James M. Poterba, Venti, and all employees in 1982. Any em- counts are also tax deductlble and I have concluded that contributions

NBER ReporterFall 1995 13. to these accounts represent new nandal assets to control for "indi- contributing begins to contribute, .) saving in large part. vidual-spedfic" saving effects. The that household should reduce non- In determining the effect on sav- analysis accounts for the explicit IRA saving. Likewise, when a ing of lRAs and 401(k)s, the key limit on IRA contributions and pla- household that was contributing problem is saver heterogeneity: ces substantial emphasis on the stops contributing, non-IRA saving some people save and others don't, change in non-IRA saving after the should increase. But when the and the savers tend to save more in IRA limit is reached. The first re- same families are tracked over all forms. For example, families suIts using this approach were time, there is little change in other with lRAs have larger balances in based on early data from the 1983 fmancial asset saving when families all financial assets than families Survey of Consumer Finances. 1 begin to contribute to an IRA (or without lRAs. But this does not Subsequent analysis was based on when they stop contributing). The necessarily mean that lRAs explain the 1980-5 Consumer Expenditure SIPP panel data allow calculation the difference. Thus a continuing Surveys and the 1984 panel of the of the change in non-IRA saving goal of the analysis has been to Survey of Income and Program Par- when IRA contributor status chan- consider different methods of con- ticipation (SIPp). 2 ges. This simple calculation con- trolling for heterogeneity, or indi- The results suggested that the troIs directly for changes in saving vidual-specific saving effects. The majority of IRA saving, even at the behavior across families since it is methods that could be used when outset of the program, represented based on changes over time for the each analysis was conducted were net new saving, and was not ac- same families. We considered the largely dependent on the data companied by large-scale reduction change in other (non-IRA) saving available at that time. As new data in other finandal asset saving. Ac- between 1984 and 1985 by IRA became available, we used altema- cording to these fmdings, increases contributor status.3 tive and possibly more robust in the IRA limits would lead to These data reveal little substitu- methods. substantial increases in IRA saving tion. The key information in this and very little reduction in other approach is the change in other fi- saving. If the IRA limit were raised, nancial assets when families began Early Parametric one-half to two-thirds of the in- Analysis of to contribute to an IRA. In particu- crease in IRA saving would be lar, we consider the change for .' Substitution at the funded by a decrease in current families that did not contribute in Outset of the IRA consumption, and about one-third 1984 but did contribute in 1985. If Program by reduced taxes; only a very small the new IRA saving substituted for proportion-at most 20 percent- other financial asset saving, one When we began this work in the would come from other saving. mid-1980s, data were available for would expect other saving to de- a limited time period; assets typi- cline between 1984 and 1985. But cally could be measured at only Following Individuals the non-IRA finandal asset saving two points in time, one year apart. Over Time at the of these new contributors declined To use these data, Venti and I de- Outset of the rnA by only $193 between 1984 and veloped an econometric model for Program 1985. This decline in other saving estimating the relationship between is only a small fraction of the in- IRA saving and other saving. With- As new data were released it be- crease in saving from the typical in a framework that allowed for came possible to use different family IRA contribution, about any degree of substitution between methods to control for heterogene- $2,300. Thus these data also sug- IRA and non-IRA saving, we asked: ity. We used two approaches based gest that even near the outset of after controlling for age, income, on traditional panel data. the IRA program there was only a small reduction il'l non-IRA saving other personal characteristics, and Changes in Other Saving When when IRA contributions began. accumulated housing and fmancial Individual IRA Participation assets, do persons who save more Status Changes Following Households in the Survey in the form of lRAs in a particular of Consumer Finances year save less in other financial as- If non-IRA saving is reduced set forms? Given age and income, when IRA saving is increased, then Using 1983 and 1986 Survey of this approach used accumulated fi- when a household that was not Consumer Finances data, it is possi- .'iF

14. NBER Reporter Fall 1995 ble to compare the asset balances 1991. Random samples of family The 401(k) Eligibility of the same households over time. types are similar in each of these "Experiment" Venti and I considered how the as­ years, but the 1984 sample has had sets of IRA contributors changed only about two years (1982 to Another approach relies on the over this time period.4 Households 1984) to accumulate 401(k) and "experiment" that is provided by that made IRA contributions over IRA balances, while the 1987 I)am­ the largely exogenous determina­ this period began the period with a pIe has had about five years, and tion of 401(k) eligibility. It consid­ median of $9,400 in other fmancial the 1991 sample about nine years. ers whether eligibility is assodated assets in 1983. Between 1983 and The central question is whether with higher levels of total saving, 1986, the median IRA assets of longer exposure to these plans re­ holding income and other demo­ these families increased from sults in higher levels of saving by graphic characteristics constant. In $1,000 to $7,000. At the same time, families who partidpate in the pro­ this case the key question is: did the median of other financial assets grams. The key test is whether families who were eligible for a increased from $9,400 to $13,500. non-lRA-401(k) assets declined as 401(k) in a given year have larger These families ended the period IRA and 401(k)"assets increased. total financial asset balances than with median total financial assets, The answer is ty{>ically no. In each families who were not eligible, or, including lRAs, of $24,000--an in­ of the six different IRA or 401(k) equivalently, did non-401(k) finan­ crease of 100 percent over assets in participant saver groups, there was cial assets decline enough to offset 1983. We determined that the in­ an increase in total financial assets the 401(k) contributions of eligible crease could not be accounted for by between 1984 and 1991, but little families?7 change in age, income, or rate of change in non-IRA or non-401(k) The inferences about the net return. Thus we concluded that it financial assets. saving effect of 401(k) contribu­ was unlikely that the IRA contribu­ It is sometimes suggested that tions depend on the similarity of tions simply substituted for saving these programs may affect house­ the saving behavior of families who that would have occurred anyway. holds with limited assets but have are and are not eligible for a little effect on wealthier house­ 401(k), controlling for income. It is holds. We have addressed this is­ important, for example, that the eli­ Following "Like" Saver sue by comparing the distribution gible group not be composed dis­ Groups Over Time of assets in 1984 and 1991. Again, proportionately of savers. The data we rely on the fact that by 1991 show little evidence of this type of Another way to control for het­ households had had much more erogeneity is to group families ac­ difference in saving behavior. The time to contribute to the saving most compelling evidence is for cording to their saving behavior­ programs.6 The higher levels of to­ their "taste" for saving-and to fol­ 1984. In that year the two groups tal finandal assets held by IRA and -eligibles and noneligibles-had low the progress of the saving of 401(k) participant families in 1991 about the same level of other fman­ these "like" families over time. Po­ was not limited to families with 5 cial assets. Thus, near the outset of terba, Venti, and 1 followed this large or small asset balances. Rath­ the 401(k) program, families that approach, grouping families ac­ er, the effect was evident across the were newly eligible for a 401(k) cording to whether they did or did entire distribution of households, exhibited about the same previous not contribute to an IRA. We also from those with the least to those saving behavior as families that did grouped families according to with the greatest assets. On the not become eligible. whether they were eligible for a other hand, across the entire distri­ 401(k) plan and whether they con­ bution, there was almost no Data for families with incomes tributed to that plan. Altogether, we change-between 1984 and 1991- between $40,000 and $50,000 illus­ considered eight different groups in the non-lRA-401(k) assets of con­ trate the findings: In 1984, newly of "saver types." We focused on tributors. At all points in the dis­ eligible and ineligible 401(k) fami­ the change in the other saving of tribution there was a fall over time lies had almost identical non- families in these groups who had in the assets of noncontributors. 401(k)-IRA assets of $5,027 and been exposed to 401(k) and IRA $5,082, respectively. By 1991, how­ plans for different time periods. ever, the median of total financial • For example, data from the SIPP assets of eligibles families was , ,:, were available for 1984, 1987, and $14,470, compared to $6,206 for in-

NBER RepotterFall 1995 15. .·-r· ..

eligible families. But in 1991, the 65 in 1991 had had nine years to tained this age in 1991 had median non-lRA-401(k) assets of the two contribute.8 total financial assets of $45,019. groups were still about the s'ame, Venti and I find that younger The median level of personal re­ $4,724 for eligible and $4,250 for tirement plan assets of the families • families had consistently larger to­ the ineligible group. If families re­ tal fmancial assets than older fami­ that reached this age in 1984 was duced saving in other forms when lies. The larger assets of the young­ $7,575. Those that reached this age they became eligible for a 401(k) er cohort are accounted for almost in 1991 had $21,613. In contrast, the other final)cial assets of these plan, the typical eligible family in entirely by more assets in IRA and 1991 would have accumulated less 401(k) plans. On average, there is families were about the same in wealth in other financial assets no difference between the other fi­ 1984 and 1991 ($19,358 and $17,950, respectively). Thus there than the typical noneligible family. nancial assets of the older and is little evidence of substitution of This was not the case. Looking younger cohorts. These results can personal retirement saving for oth­ over all income intervals, the total be illustrated by comparing the as­ er fmancial assets. In contrast, the fmancial assets of the eligible fami­ sets of families who reached ages financial assets of families that at­ lies by 1991 were typically two to 60 to 64 in 1984 with the assets of tained age 60 to 64 in 1991 but did eight times as large as the total fi­ families that attained that age in not participate in the personal re­ nancial assets of the noneligible 1991. tirement plans, were somewhat families. Thus this approach shows To control for heterogeneity, the lower than similar families who no substitution of 401(k) contribu­ data for all families combined­ reached this age in 1984. tions for other financial asset saving. both contributors and noncontribu­ This is the typical pattern for all Indeed, for all income groups, tors-are the most compelling. The ages that we considered, between eligible households have greater mean of total financial assets of all 45 and 70. Indeed, the analysis total financial assets than ineligible families that attained age 60 to 64 suggests that if current patterns households at virtually all pOints in 1984 was $32,807; the mean of persist, families who reach retire­ across the entire distribution of fi­ those who attained this age in 1991 ment age 25 or 30 years from now nancial assets. But there is virtually was $39,105 (in 1991 dollars and will have much more in financial no difference across the entire dis­ controlling for income, age, educa­ assets (at least $30,000) than fami­ tribution of the other fmancial as­ tion, and marital status). The in­ lies currently attaining retirement • sets of eligible and ineligible crease was accounted for almost age; the difference will be due households. entirely by personal retirement sav­ solely to assets in personal retire­ ing-$4,027 for the cohort that at­ ment accounts. tained ages 60 to 64 in 1984 com­ Cohorts and the pared to $10,995 for the cohort that In Canada Effects of Retirement attained this age in 1991. There Registered Retirement Saving Saving Programs was essentially no cohort differ­ Plans (RRSPs) were first introduced ence in other financial assets In the United States in Canada in 1957. Like contribu­ ($28,780 for the older cohort and tions to an IRA in the United No single method provides a $28,110 for the younger cohort). States, contributions to an RRSP are perfect control for all possible Because fewer than half of families deductible from income for tax forms of heterogeneity. 1 believe participated in these programs, the purposes. Interest accrues tax free that the method that comes the median for all families is zero and until withdrawal, when taxes are closest is cohort analysis. Essential­ therefore is not informative. The paid. The contribution limits were ly, this method compares the assets data for families who participated increased substantially in the early of persons who are the same ex­ in personal retirement saving plans 1970s, and RRSPs were promoted cept for age. Hence some groups­ provide a better measure of the widely. Since then, they have be­ cohorts-had longer to contribute potential of the plans to augment come a very prominent form of to special saving programs. For ex­ the financial assets of retirees: the saving. Annual contributions grew ample, families that reached age 65 median level of total personal fi­ from $225 million in 1970 to almost in 1984 had had only two years to nancial assets of contributor fami­ $3.7 billion in 1980 to $16 billion contribute to an IRA or to a 401(k) lies that attained age 60 to 64 in by 1992, when they accounted for plan. But families who attained age 1984 was $29,847. Families that at- about one-third of aggregate per- • 16. NBER Reporter Fall 1995 • sonal saving. In 1992 about 33 per­ whose aftertax rate of return was 2S. F. Venti and D. A. Wise, "Have IRAs " cent of families contributed an av­ unaffected by the legislation was Increased U.S. Saving? Evidence from erage of $4,180 to RRSPs. Now reduced by 40 to 50 percent. High­ Consumer Expenditure Surveys, " Quar­ RRSP contributions exceed the total er-income families that lost the up­ terly Journal of Economics (August 1990), pp. 661-698; and "Tbe Saving of employee and employer contri­ front deduction virtually stopped Effect of Tax-Deferred Retirement Ac­ butions to employer-provided pen­ contributing, even though the re­ counts: Evidence from SIPp, " in N ation­ sion plans. turn still accrued tax free. These re­ a1 Saving and Economic Performance, Based largely on "cohort" analy­ sults suggest that the promotion of B. D. Bernheim and]. B. Shoven, eds. saving plans may have an impor­ Chicago: University of Chicago Press, sis like the procedure illustrated 1991. tant effect on their use. The results above, Venti and I conclude that 3S. F. Venti and D. A. Wise, "Individual the data taken as a whole suggest also suggest that the up-front tax Response to Retirement Savings Pro­ that RRSPs have contributed sub­ deduction may have been a critical grams: Results from u.s. Panel Data, " stantially to personal saving in Can­ determinant of the decision to Richerche Economiche, special issue, ada. 9 In virtually no case do ,the make an IRA contribution. forthcoming. Although the IRA participation 4S. F. Venti and D. A. Wise, "Govern­ data suggest substitution of RRSPs ment Policy and Personal Retirement for other forms of retirement sav­ rate was never more than 16 or 17 Saving, " NBER Reprint No. 1752, Octo­ ing. In the two decades prior to the percent, the participation rate ber 1992, and in Tax Policy and the growth in popularity of RRSPs, the among persons eligible for a 401(k) Economy; Volume 6,]. M. Poterba, ed. personal saving rate in Canada was is above 60 percent, and is at least Cambridge: MIT Press, 1992, pp. 1-41. typically below the U.S. personal 50 percent even for low-wage earn­ 5]. M. Poterba, S. F. Venti, and D. A. ers. In addition to greater financial Wise, "401(k) Plans and Tax-Deferred saving rate. Since that time, the Saving, " in Studies in the Economics of personal saving rate in Canada has incentives provided by employer Aging, D. A. Wise, ed. Chicago: Univer­ become much higher than in the matching rates, a likely explanation sity of Chicago Press, 1994; and 'Vo United States. Although it is diffi- is that contributions are typically 401(k) Contributions Crowd Out Other . cult to make judgments about the by payroll deduction, and thus Personal Saving?" Journal of Public Economics, forthcoming. ( ..• RRSP effect on saving based only sheltered from current expenditure urges: the "out of sight, out of 6]. M. Poterba, S. F. Venti, and D. A. ~. on the trends in U.S. and Canadian Wise, "Targeted Retirement Saving and aggregate saving rates, a large frac­ mind" feature of the payroll deduc­ the Net Worth of Elderly Americans," tion of the current difference can tion arrangement may provide American Economic Review 84, 2 (May be accounted for by RRSP saving. needed self control. More general­ 1994). ly, the fmdings bring into question 7Poterba, Venti, and I also used this ap­ the extent to which different forms proach in "401 (k) Plans and Tax-De­ of saving are treated as perfect ferred Saving" and "Do 401(k) Contri­ To Consider butions Crowd Out Other Personal Sav­ substitutes by real people. Thus in ing?" op. cit. This series of papers suggests my view the results provide moti­ 8Tbis is the approach that Venti and I that the IRA and the 401(k) pro­ vation to look more broadly for fol/owed in our analysis of "Tbe Wealth grams have had very important ef­ explanations of saving behavior. of Cohorts: Retirement Saving and the fects on saving and are likely to Changing Assets of Older Americans, " add substantially to the assets of fu­ NBER Working Paper No. 4600, Decem­ ture retirees. In addition, I believe ber 1993. that the research has another fun­ 9S.F. VentiandD.A. Wise, "RRSPsand Saving in Canada, "mimeo, 1995. damental underlying theme: stan­ dard assumptions about the deter­ IS. F. Venti and D. A. Wise, "Tax-De­ minants of saving behavior leave ferred Accounts, Constrained Choice, and Estimation of Individual Saving, " important aspects of actual saving Review of Economic Studies 53 unexplained. For example, the Tax (1986); "IRAs and Saving, " in The Ef­ Reform Act of 1986' led to reduc­ fects of Taxation on Capital Accumula­ tions in IRA contributions well be­ tion, M. Feldstein, ed. Chicago: Univer­ yond the effects that might have sity of Chicago Press, 1987; and D. A. Wise, "Individual Retirement Accounts been predicted by models that rely and Saving," in Taxes and Capital For­ • on the aftertax rate of return. Even mation, M. Feldstein, ed. Chicago: Uni­ (Co' the contribution rate of persons versity of Chicago Press, 1986.

NBER Reponer Fall 1995 17. NBER Profile: Anil K Kasbyap

Anil K Kashyap is an NBER fac­ nels, and the sources of business ulty research fellow and an associ­ cycles. His work has been pub­ ate professor of business econom­ lished in a number of academic ics at the University of Chicago's journals and books. With Takeo Graduate School of Business .. He is Hoshi, he currently is working on a a member of the NBER Programs book titled Keiretsu Financing, in Corporate Finance and Monetary which will be published by the Economics and serves as the head MIT Press next year. of an NBER working group that Kashyap is also a member of the studies the Japanese economy. University of Chicago's Center for Kashyap earned his Ph.D. in East Asian Studies, an associate fel­ economics in 1989 from MIT, and low in the Federal Reserve Bank of his bachelor's degree in economics Chicago's Research Department, and statistics from the University of California, Davis, in 1982. He cur­ and a member of the Advisory rentl y teaches courses on "Money Council to the Japanese Ministry of and Banking" and "An Introduction International Trade and Industry's to the Japanese Economy." study of market access. He previ­ Kashyap's principal research in­ ously served as a staff economist for the Board of Governors of the 1 terests include the changing nature who is 2 / 2. His hobbies are "rotis­ of the Japanese financial system, Federal Reserve System. serie baseball," bridge, and golf. the impact of monetary policy be­ Kashyap and his wife Katie Mer­ He is also a season ticket holder of yond standard interest rate chan- rell have one daughter, Laurie, the Chicago Cubs~

NBER Profile:Jeremy C. Stein

Jeremy C. Stein has been an the Council of Economic Advisers. NBER research associate in the cor­ Returning to Cambridge, Stein porate finance program since 1992. became an associate professor at He is also the J. c. Penney Profes­ MIT in 1990, a professor of finance sor of Management at MIT's Sloan there in 1993, and the J. c. Penney School of Management. Professor in 1994. Stein received his A.B. in eco­ In addition to his teaching and noinics from Princeton University research, Stein serves as associate in 1983 and his Ph.D. in economics editor of the Quarterly Journal of from MIT in 1986. From 1986--7 he Economics and the Journal of Fi­ was an intern at Goldman Sachs & nance. He is also on the editorial Co. and a research fellow at the board of the American Economic Harvard Business School (HBS). He Review. was an assistant professor of fi­ Stein is married to Anne Maas­ nance at HBS from 1987-9, when land. They live in Lexington (MA) he went to Washington to spend a with their three children: Carolyn, 1 e. year as senior staff economist at 4 / 2; Alison, 2; and Jason, 5 months. Jt

18. NBER RejJorterFall 1995 ( t NBER Profile: Richard H. Thaler

Richard H. Thaler is a professor also has been a visitor at the Uni­ of behavioral science and econom­ versity of British Columbia, MIT, ics, and director of the Center for and the Russell Sage Foundation. Decision Research, at the Universi­ Along with other books and ty of Chicago's Graduate School of journal articles, Thaler wrote a Business. He is a member of the quarterly column for several years NBER's Program on Asset Pricing, in the Journal of Economic Perspec­ and also codirects (with Robert J. tives on anomalies. The first set of Shiller) the Bureau's Project on Be­ these columns was published in havioral Economics. He has been an book form by Princeton University NBER research associate since 1990. Press under the title The Winner's Thaler received his Ph.D. in Curse: Paradoxes and Anomalies of 1974 from the University of Roch­ Economic Life. He recently re­ ester, where he also taught for a sumed writing this column. few years. His research in behav­ Thaler and his "significantly bet­ ioral economics really began dur­ ter," France Leclerc, a marketing ing a year he spent at the NBER's professor at the University of Chi­ Palo Alto office in 1977-8. After cago, live in the Lincoln Park sec­ fine wine cool in the summer, they that, he joined Cornell's Johnson tion of that city. In addition to try­ enjoy travel-especially to places Graduate School of Management, ing to keep themselves warm in with good restaurants, scuba div­ (t where he remained until 1995. He the winter and their collection of ing, or skiing. Conferences

NBER Reporter Fall 1995 19. t \

Using a unique dataset covering responsibility for providing jobs for lady young workers-left their jobs •.. individuals in 23 countries over an those who want them, and for con­ when their wage fell below the ' eight-year period, Blanchflower trolling both wages and prices. 5). minimum. In the United States, a fInds that a number of traits distin­ If unemployed, the young are es­ comparable effect released individ- guish the young from the older: 1) pecially likely to say how impor­ uals into employment as their real They have a relatively high proba­ tant they believe work is. wages passed a declining real min- bility of being unemployed and a BonnaI, Fougere, and Lollivi­ imum wage. relatively low probability of being er present empirical evidence on Gelot and Osbert point out that a member of the labor force. Youth how becoming employed depends over the past 20 years, youth un­ unemployment rates are especially on individuals' past employment employment became the most no­ high in Bulgaria, Israel, Ireland, Po­ histories. This stylized fact, charac­ ticeable fact in the French labor land, and Slovenia. 2) They have a terizing the youth labor market in market. Since the 1970s, public in­ relatively low probability of being France, suggests that employers tervention has changed the process self-employed or a member of a rank and select unemployed young of becoming employed for youth, trade union. In surveys, the young workers on the basis of their past and particularly for those young are particularly likely to say that occurrences of employment and people who are least qualifIed, by they would like to be self-em­ unemployment, rather than on the broadening the role of public em­ ployed if they had the choice; they basis of how long their current un­ ployment and vocational training. are less likely than others to report employment spell is. The Individualized Training Credit, that trade unions in their country Abowd, Kramarz, Lemieux, set up in the fall of 1989, makes this have too much power, and are and Margolis fInd that both the new approach concrete, by extend­ more likely to say that trade unions French and the American minimum ing access to skill development to are good for the country as a wages are associated with relatively the most disadvantaged young peo­ whole. 3) The young report being important effe.cts on employment. ple between the ages of 16 and 25. unusually happy with their lives. 4) As the French minimum wage in­ Requests for copies of these pa­ They are especially likely to be­ creased from 1981 to 1989, more pers should be addressed to the lieve that the government has the and more French workers-particu- NBER's Conference Department. •

• 20. NBER Reporter Fall 1995 The United States has ratified a dataset based on line-item tariff AD law by U.S. firms has the pecu­ both NAFTA and the Uruguay codes, he examines the trade pat­ liar side-effect of benefiting non­ • Round, which together include terns of both countries named in named countries that are active in (. sweeping trade liberalization mea­ the petitions and countries not sub­ the areas of investigation. sures. Simultaneously, those oppos­ ject to investigation. He finds first Elliott and Richardson investi­ ing trade liberalization or seeking that AD cases are aimed primarily gate the determinants of success new trade protection have battled at restricting large competitors. Sec­ and failure of Section 301 actions, with extraordinary intensity and ond, AD duties substantially restrict extending the work of Bayard and some success. Cumby and Moran the volume of trade from named Elliott (1994) by including an addi­ examine one of the most vigorous­ countries, espedally for cases with tional 15 cases to the 72 previously ly contested areas in the Uruguay high duties. Third, AD actions that considered. They conclude, among Round, the case of establishing a are rejected still have an important other things, that the Bayard-Elliott framework for antidumping actions impact on named country trade, es­ findings are still relevant for the in the World Trade Organization. pecially during the period of inves­ updated and expanded sample of Here efforts at trade liberalization tigation. Fourth, there is a substan­ cases. They also conclude that suc­ were not successful. In fact, there tial trade diversion from named to cessful 301 cases are associated was an extension and codification nonnamed countries, and the diver­ with a high share of GNP exported of a new protectionist regulatory sion is greater as the estimated to the United States and a low U.S. regime worldwide, leavened only duty grows larger. Because of the trade balance with the target coun­ by some modest reforms of existing diversion of imports, the overall try. Section 301 cases about prac­ procedures. The authors ask how this volume of trade continues to grow, tices that make unfair distinctions campaign on behalf of antidump­ even for those cases that result in at the border are far more success­ ing protectionism succeeded in the duties. Fifth, despite the diversion ful than those about nonborder midst of a major tide flowing in the of imports, AD law still has impor­ practices. By contrast, there is little direction of liberalization, what it tant effects, because it induces sub­ difference in success rates between says about the process, and what it stantial import price increases both cases involving taxes and subsidies portends for trade battles to come. by named and by nonnamed coun­ and those involving less transpar­ , Prusa analyzes the effectiveness tries. Finally, because of the diver­ ent regulatory and nontax prac­ (\oct of antidumping (AD) actions. Using sion in imports, aggressive use of tices, nor between cases involving

NBER Reporter Fall 1995 21. traditional trade concerns and short tun; while realizing that their stein argues, fairly conventional in- those involving less well-estab- choices of capacity and yield-im- terventions generated the high lev- lished "new issues" (induding proving Rand D in the medium els of corporate ownership in Ja- trade in services and protection of and long run will affect market pan. It is these stable owners, and • intellectual property). There also is price. This model fits the actual not keiretsu, that make takeovers little difference between success configuration of the flat panel in- difficult in Japan, he fmds. rates in agricultural versus nonagri- dustry in 1994. Policy simulations Dick explains how deregulation cultural cases, nor among various show that steady-state yields and policies led to the telecommunica- trading partners. profits are lower, while prices are tions industry's bilateral (U.S.!Ja- Levinsohn explains that the re- higher, with subsidies for capacity pan) trade imbalance. He con- cent trade frictions between the acquisition rather than with Rand dudes first that the timing and United States and Japan in the auto D subsidies. This is because a scope of the trade imbalance coin- parts market are based on these firm's incentives to undertake R cided with major deregulatory ac- facts: Initially, there was not much and D are diminished by a subsidy tions in the United States; Japanese trade because Japanese cars were on capacity costs. policy alternatively played neutral made in Japan with mostly Japa- Goldberg and Knetter study and reinforcing roles. Second, the nese parts; the same was generally the causes and effects of the Export impact of deregulation on trade true for North American cars. Most Enhancement Program (EEP) for ranged widely within the industry, trade consisted of U.S. imports of wheat. They find that the over- Dick finds. Monopolies in terminal Japanese parts, which contributed whelming causes of the EEP-fal- equipment, which had been sus- to a growing bilateral trade deficit tering export markets and swelling tained only by regulatory barriers in auto parts. In an effort to ad- government stocks-are primarily to entry, were eroded quickly by dress this deficit, the United States attributable to the overvaluation of differentials in international factor threatened tariffs on 13 models of the dollar in the 1980s, not to the costs once regulation was lifted. Japanese cars. This might seem like increase in EC subsidies to wheat Monopolies in network equipment, a very indirect way to address the farmers in 1985. They also fmd that which were sustained additionally parts trade, but more direct ave- the EEP has done little to restore by economic barriers to entry, nues were either ineffective or too export market shares since 1985 were eroded only when deregula- costly. The threatened tariffs would beyond what might have been ex- tion coincided with major advances • have resulted in drastically r~duced pected as a result of the dollar's in technology. sales of the 13 models, and total depreciation. Furthermore, the EEP Blonigen and Feenstra analyze Japanese profits would have fallen does not appear to have increased a panel dataset of four-digit SIC- around 12.5 percent. The European U.S. domestic wheat prices relative level observations of Japanese for- firms would have captured many to region-specific export prices. eign direct investment (FDD in man- of the lost Japanese sales, and U.S. Weinstein suggests that it is not ufacturing into the United States in firms would have been pretty cultural or institutional factors that the 1980s. They find that higher much unaffected by the tariffs. The produce keiretsu, or corporate threats of protection lead to greater result was an unenforced trade groups, in Japan, but rather con- FDI flows. A rise in the expected pact, in which Japan agreed to buy ventional government policy. In probability of protection from 5 to substantially more U.S. parts, and Japanese industrial policy, corpo- 10 percent means over a 30 per- the United States agreed to drop rate groups form as a result of mul- cent increase in next-period FDI the threatened tariffs. Thus there is tibillion- and multitrillion-yen sub- flows for an average industry, they a trade promotion pact, although it sidy programs for politically power- estimate. resulted from a threat of trade ful constituencies. Thus Weinstein The Canada-U.S. Free Trade protection. concludes that we should reexam- Agreement of 1988, while receiving Krishna and Thursby examine ine the notion that conventional much less attention in the United the possible consequences of sub- policies are not important. Further, States than the follow-up agree- sidies to Rand D and to volume the data suggest that levels of for- ment that included Mexico, drew production proposed under the eign direct investment into Japan adamant opposition in Canada. In Clinton administration's flat panel are not nearly as out of line with the early 1990s, critics attributed display initiative. In their model, international levels as is believed the alleged loss of 350,000 Canadi- firms behave competitively in the widely. More importantly, Wein- an manufacturing jobs to the elimi- • 22. NBER Reporter Fall 1995 .:,a nation of tariffs between the coun­ suggest that transplant production ue added in foreign assembly :fI tries. Head and Reis examine may reduce the U.S. automotive plants in the six corresponding changes in Canada's shipments rel­ deficit with Japan, but will not Mexican border cities, Hanson ative to U.S. shipments in 129 eliminate it. finds that the growth of export as­ matched manufacturing industries. Hanson examines how the sembly in Mexico increases the de­ They find that most Canadian in­ growth of offshore assembly in mand for manufactured goods pro­ dustries experienced relative Mexico has affected manufacturing duced in U.S. border cities. The ex­ declines from 1987 to 1992. The activity in U.S. border cities. Under pansion of export-assembly plants elimination of Canadian protection the offshore assembly provision of in Mexican border cities explains appears to be related to the de­ over half of the increase in dura­ clines; however, a deeper reces­ the U.S. tariff schedule, goods as­ ble-goods activities in U.S. border sion in Canada also contributed. sembled abroad with U.S.-manufac­ tured components receive prefer­ cities from 1975-89. Swenson studies the domestic ential tariff treatment upon reentry Also participating in this confer­ content decisions of automakers in the the United States between 1984 and into United States. The majority ence were: Anne o. Krueger, NBER 1993, using foreign trade zone ac­ of foreign assembly plants in Mexi­ and Stanford University; Robert E. tivity to observe individual sourcing co, most of which are owned by Lipsey, NBER and Queens College; and production. She shows that al­ U.S.-based multinationals, are con­ and Suddhasatwa Roy, Purdue Uni­ though the domestic content of centrated along the border with the versity. The papers presented, and Japanese firms is rising, differences United States. After combining data their discussions, will be published are not being eliminated complete­ on employment and earnings in by the University of Chicago Press. ly. Also, the apparent elasticity of two-digit manufacturing industries The release of this volume will be substitution is lower for Japanese for the six largest U.S. border cities announced in an upcoming issue than for U.S. firms. These results with data on employment and val- of the Reporter.

NBER Reporter Fall 1995 23. Deardorff derives equations for tional trade well into the postwar that rich and poor countries have the value of bilateral trade from period, although the size and sig­ in choosing their partners in trade two extreme cases of the Heck­ nificance of the coefficient on arrangements; and 3) whether the scher-Ohlin Model. The first case lagged trade declines with the pas­ welfare consequences of continen­ is free trade, in which the absence sage of time. Eichengreen and Ir­ tal preferential trade arrangements of all barriers to trade in homoge­ win conclude that the standard for­ depend on relative endowments. neous products causes producers mulation, which omits a role for Whalley argues that regional and consumers to be indifferent historical trade flows, may be bi­ trade arrangements around the among trading partners. Resolving ased. Because there are reasons to world are surprisingly different this indifference randomly, expect­ anticipate a positive correlation be­ from one another, in part because ed trade flows correspond exactly tween the predominant direction of countries of different size are in­ to the simple frictionless gravity trade flows in the past and mem­ volved, and in part because they equation if preferences are identi­ bership in preferential arrange­ have different objectives. Some cal and homothetic, or if demands ments in the present, there may be agreements have strategic under­ are uncorrelated with supplies. In a tendency to spuriously attribute pinnings, and implicitly form part the second case, countries produce to preferential arrangements the ef­ of security arrangements (as in Eu- .'" distinct goods, either with com­ fects of historical factors, and to ex­ rope). Some are the outcome of plete specialization or with a vari­ aggerate the influence of the former. smaller countries seeking to obtain ety of other models. The main les­ Kowalczyk and Davis discuss more security in their access to sons from the paper are first, that it tariff phaseouts from theoretical, larger country markets (as in the is not difficult to justify even sim­ empirical, and historical perspec­ Canada-U.S. agreement). Some re- ple forms of the gravity equation tives. They find that high U.S. tar­ flect agreements used by smaller from standard trade theories; and iffs and little intraindustry- trade are countries to underpin their com­ second, that because the gravity associated with long NAFTA phase­ mitment to domestic policy reform equation appears to characterize a out periods for U.S. imports from (Mexico in NAFTA). Others reflect large class of models, its use for _ Mexico. Mexico's phaseouts are tactical considerations, conscious empirical tests of any of them is correlated with those of the United efforts to use prior regional agree­ suspect. States but not generally with Mexi­ ments to influence subsequent Eichengreen and Irwin investi­ co's tariffs. multilateral negotiation (services in gate the degree to which history (in Spilimbergo and Stein present Canada-United States and in terms of past trade flows) continues a model in which trade is motivat­ NAFTA). Some are stepping-stones to influence current trade flows. ed both by preference for variety to eventual deeper integration (the They analyze data for the immedi­ and by Comparative advantage. EEC in the 1960s), while others are ate postwar period in order to con­ They use this framework to analyze consciously stand-alone, more shal- centrate on the effects of interwar the welfare implications of trading low agreements (NAFTA). Whalley trade patterns for the postwar de­ blocs among countries with differ­ stresses the need to recognize the velopment of trade, including the ent endowments and with and objectives countries get for their in­ effects of postwar regional and glo­ without transportation costs. They volvement in regional trade agree- bal liberalization initiatives, such as address the following issues: 1) the ments when evaluating the impacts the EEC and the GATT. They find welfare implications of the consoli­ of any particular agreement. that prewar trade patterns continue dation of the world into a few trad­ Using the gravity model of bilat- • to shape the direction of intern a- ing blocs; 2) the different incentives eral trade and an updated dataset _

24. NBER Reporter Fall 1995 \

\

covering 1970-92, Frankel and vorable or unfavorable political also find that distance can explain Wei map out the current pattern of economy effects of regionalism are failures of LOOP, suggesting that regionalization in trade. They also likely to· dominate. Their tentative failures arise from imperfect market estimate the role of currency links verdict is favorable: a majority of integration. However, these sources within some major country group­ free trade agreements, such as ASEAN do not explain all of the failure of ings in promoting intragroup trade. and Mercosur, have increased trade LOOP. They speculate that integrat­ Then, they review various political with nonmembers, even while they ed marketing and distribution sys­ economy arguments that others have concentrated trade dispropor­ tems within regibns cause LOOP to have made regarding regionalism: tionately with each other. hold more nearly intraregionally. for example, that it can help build Engel and Rogers find that the Also participating in this confer­ political momentum for multilateral law of one price (LOOP) holds ence were: NBER President Martin liberalization, or that it can under­ more nearly for country pairs that Feldstein, also of Harvard Universi­ mine more general liberalization. are within geographic regions than ty; Anne O. Krueger, NBER and They present a simple model illus­ for country pairs that are not. Using Stanford University; and Sang­ trating one possible beneficial ef­ disaggregated consumer price data Seung Yi, Dartmouth College. The fect of trade blocs in building fur­ from 23 countries, including data papers presented, and their discus­ ther trade liberalization. The result from eight North American cities, sions, will be published by the Uni­ could go the other way just as easi­ they fmd that f:Ulures of LOOP are versity of Chicago Press. The re­ ly, however. Frankel and Wei re­ related closely to nominal ex­ lease of this volume will be an­ turn to the gravity model estimates change rate variability, suggesting a nounced in an upcoming issue of to tentatively assess whether the fa- link to sticky nominal prices. They the Reporter.

Fullerton discussed environ­ ton showed that these compliance the retirement component of Social mental taxes, and some flaws in costs are high relative to the small Security, which he estimates can their current design. Each environ­ amount of revenue raised from generate very major increases in mental tax in the United States is each tax. In addition, current U.S. output and living standards in the designed to collect revenue for a environmental tax burdens are long run. These gains would come trust fund used to clean up a par­ passed from the taxed industries to at the expense of existing genera­ ticular pollution problem. Since all other industries, he demonstrat­ tions, but not ex!=lusively at their there is a different tax for each ed. Thus the added cost of admin­ expense. Indeed, Kotlikoff showed trust fund, each tax rate is typically istering each separate tax neither that after existing generations have less than 1 percent. But there is achieves the targeted incentives been compensated fully for their also an administration and compli­ nor lessens the targeted burdens. losses from privatization, future ance cost involved, since taxpayers Kotlikoff considered the issue generations will be better off. must read multiple sets of rules of privatizing Social Security. He When the initial tax structure fea­ and fill out multiple forms. Fuller- analyzed a proposal for privatizing tures a progressive income tax, the

NBER Reporter Fall 1995 25. linkage between benefits and taxes consumption taxes. The destina­ because of induced changes in is low, a consumption tax is used tion-basis VAT and national sales work, in the form of compensation, .'! to finance Social Security benefits tax require a one-time appreciation and in tax-deductible expenditures. during the transition to privatiza­ of the dollar, a· shock that other Eissa described her most recent tion, and existing generations are consumption taxes avoid. General­ work on the labor supply effects of compensated fully for their privati­ ly, Hall pointed out, consumption tax reform. Research by Eissa and zation losses, there is still a 4.5 per­ taxes would impose a new tax on others shows that taxable income cent welfare gain for future genera­ the rental value of owner-occupied increased after the tax reforms of tions, Kotlikoff estimated. If these housing, which might have a small the 1980s, as did wages and sala­ circumstances don't hold, then the adverse net effect on housing. This ries. Female labor supply increased efficiency gains from privatization effect could be more than offset by as a result of these changes. But are likely to be smaller, and possi­ the introduction of a personal what about men, who remain the bly even negative. housing deduction, he claimed. primary earners? Eissa finds that Hall discussed how various tax Feldstein and Feenberg ana­ male participation in the labor reforms might affect relative prices. lyze taxpayer behavior after the force and total hours worked re­ All forms of consumption taxes 1993 increase in personal tax rates sponded only slightly to the tax re­ have the same basic, desirable ef­ using data from recently released forms, mostly after the Tax Reform fect in the economy: they eliminate '93 tax returns. Their preliminary Act of 1986 (TRA86). For example, the distortion present in an income analysis confirms that taxable in­ total hours worked by men with tax between current and future come is quite sensitive to marginal more than 16 years of education consumption. But different ways of rates. As a result, the '93 tax rate increased by approximately 1 to 2 administering consumption taxes increases raised only about one­ percent on average following have different effects in the long third the revenue that would have TRA86, she fmds. run, and especially during the trans­ been raised with no behavioral re­ The proceedings of this confer- ition, 'Hall explained. For example, sponse. Further, Feldstein and Feen­ ence will be published in paper­ sales taxes and value-added taxes berg estimate that for every dollar back by the MIT Press; the volume (VATs) require a one-time increase of additional revenue collected by should be available in about six in the price level, unlike the Hall­ the government, taxpayers incurred months. Individual papers can be Rabushka flat tax and personal a deadweight loss of three dollars ordered directly from the NBER. •

Bureau News

• 26. NBER Reporter Fall 1995 Lisa Lynch around the world attended the Commission to Study the Consu- :- NBER's 17th annual Summer Insti- mer Price Index" who are affiIiated Becomes Chief tute. This year's program was fund- with the NBER are: Robert J. Gor- Economist of ed primarily by a grant from the don of Northwestern University, a Lynde and Harry Bradley Founda- research associate in the Bureau's Labor tion, with additional support from productivity and economic fluctua- Department the National Science Foundation tions programs; and Zvi Griliches and the National Institute on of Harvard University, a research associate and director of the Lisa M. Lynch, who has been af- Aging. The paperS presented at 27 NBER's Program on Productivity. filiated with the NBER since Sep- different sessions covered a wide Griliches, who has been interested tember 1985 and most recently was variety of topics. A list of all papers and work in progress can be in price measurement for many a research associate in the Bureau's years, and Ernst R. Berndt of MIT, labor studies program, went on obtained by writing to: Summer Institute Catalogue, NBER, 1050 organized a special session at the leave in September to become 1995 NBER Summer Institute on Chief Economist at the U.S. Depart- Massachusetts Av:~nue, Cambridge, MA 02138-5398.• "The CPI: Interpretation, Bias, In- ment of Labor. Lynch also holds the dex of Escalation, and Issues Merit- William 1. Clayton chair in Interna- ing Research." That day-long meet- tional Economic Affairs at Tufts ing brought NBER researchers to- University's Fletcher School of Law gether with representatives of the and Diplomacy. At the Department Bureau of Economic AnalYsis, Bu- of Labor, she replaces Alan B. Krue- reau of Labor Statistics, Bureau of ger, also an NBER research. asso- NBER Researchers the Census, and Federal Reserve ciate in the labor studies program, on Congressional Board to discuss the CPI. who has returned to Princeton Uni- Several of the commission's versity as a professor of economics. CPI Advisory fmdings were based on previous t Commission Bureau research, including: The Measurement of Durable Goods edited by Robert Gordon, A bipartisan congressional advi- Prices, J. Gruber Receives and published by the University of sory commission, chaired by NBER Chicago Press in 1990; The Eco- NSF Presidential Research Associate Michael J. Bos- nomics of New Goods, edited by kin, also of Stanford University, has Faculty Award Gordon and Timothy Bresnahan, calculated that in the recent past which is forthcoming from the Uni- the Consumer Price Index (CpO Jonathan Gruber, an NBER fac- versity of Chicago Press; "Valuation has overstated inflation by about ulty research fellow who teaches at of New Goods Under Perfect and one and a half percentage points MIT, recently won a National Sd- Imperfect Competition," NBER per year. It also concluded that in ence Foundation Presidential Fac- Working Paper No. 4970 by Jerry ulty Fellowship. There are 15 of the future, this bias is likely to be A. Hausman; "Axiomatic and Eco- these awards given in all of the sci- about one percentage point per nomic Approaches to Elementary ences and social sciences; Gruber year. Correcting this future bias Price Indexes" and "Price and Vol- is the only social scientist to win could reduce the federal deficit by ume Measures in a System of Na- one this year. an estimated $634 billion over the tional Accounts," NBER Working next ten years. The CPI, which is Papers No. 5104 ~nd 5103 by W. calculated by the Bureau of Labor Erwin Diewert; and several NBER Statistics, is the inflation measure working papers by Berndt and Gri- 1995 Summer that serves as the basis for cost-of- liches, alone and with coauthors. living adjustments to Social Securi- The working papers may be or- Institute ty and veterans' benefits, as well as dered from the NBER's Publications income tax rate brackets and Department; the books must be or- ) Over 800 economists from 194 exemptions. dered directly from the University .. universities and organizations Other members of the "Advisory of Chicago Press. \

NBER ReporterFall 1995 27. New Directors Elected

At the fall meeting of the NBER's Cassel Professor with respect to M.A., and Ph.D. from the Universi­ Board of Directors, three new di­ Money and Banking at the London ty of Michigan. Before coming to rectors were elected: George A. School of Economics in 1978--80. He Temple, he taught at Stanford Uni­ Akerlof, representing the University has been an NBER research asso­ versity's Graduate School of Busi­ of California, Berkeley; William C. ciate in the economic fluctuations ness and Purdue University's Kran­ Dunkelberg, representing the Na­ and monetary economics programs. nert Graduate School of Management tional Association of Business Econ­ Dunkelberg was president of Yoffie is the Max and Doris Starr omists (NABE); and David B. Yoffie, NABE in 1993-4 and currently Professor of International Business representing Harvard University. serves on its Board of Directors. Administration at Harvard Business Akerlof is a professor of eco­ He also has been chief economist School, where he has taught since nomics at Berkeley. He holds a of the National Federation of Inde­ 1981. He received his B.A. from B.A. from Yale University and a pendent Business since 1971. A Brandeis University, and his M.A. Ph.D. from MIT. He began his ca­ professor of economics at Temple and Ph.D. from Stanford University. reer as an assistant professor at University and the director of that This academic year, he is a visiting Berkeley in 1966, was promoted to school's Center for Entrepreneur­ scholar at Stanford's Graduate associate professor in 1970, and to ship, Dunkelberg holds a B.A., School of Business. professor in 1977. He was also the • has some coverage of the United data on the ICPSR tape, and we Improved Kingdom, France, and Germany, have converted the dataset to an Accessibility of although it predominantly covers accessible format. Our NBER Work­ the United States. ing Paper (No. 5186) of the same theNBER's title provides details. When the data were collected, The revised and verified version Historical Data the NBER researchers copied them of the NBER Macroeconomic His­ by hand from the original sources Daniel Feenberg, NBER, and tory Database can be accessed only onto data sheets. In 1978, the Inter­ Jeffrey A. Miron, NBER and via the Internet. Point your brows­ University Consortium for Political Boston University er at and Social Research (ICPSR) trans­ During the first several decades ferred the data from the NBER's gopher:llnber.harvard.edu of its existence, the NBER assem­ handwritten sheets to magnetic or bled an extensive dataset that cov­ tape. Unfortunately, most macro­ http://nber.harvard.edulmacrohist.html ers all aspects of the pre-World economic researchers now use PCs This is a WAIS searchable data­ War I and interwar economies, in­ rather than mainframes, so use of base, and you will be prompted for cluding production, construction, the tape is cumbersom~. Moreover, search terms. A catalogue of all series employment, money, prices, asset the process of transferring the data in the original dataset can be pur­ market transactions, foreign trade, from the NBER's handwritten sheets chased from the NBER's Publications and government activity. Many of to the ICPSR tape introduced a num­ Department. The data are stored in the series are highly disaggregated, ber of mistakes. the Micro-TSP .DB format. This for- and many exist at the monthly or We have used the NBER's origi­ mat is pure ASCII and includes all • quarterly frequency. The dataset nal handwritten sheets to verify the documentation. •

28. NBER Reporter Fall 1995 . Andre'Y B. Abel c John M: Abowci . JamesP.Adams . . Joshua Aizenman . !Alberto F: Alesina" . .;' Steven·G. Allen' .: Joseph G: Altonji ..•. Albert Ando Joshua Angrist

:;>~~~~yL$~R~J~n" '.. Sherwiil Rosen jUlio J.Rotemberg '. '. Michael 'Rothschild Richard S. Ruback

~On: Leave for Go'i~r.nmerit'Service ~ L!_:t_o_n__ Le_"_av_e __ fo_r_N __ O_n~g_ci_V_ernm_"~e_'n_t_s_e_nn __·c_e ______~ ______~ ______~

NBER ReporterFall 1995 29. Faculty Research Fellows for 1995-6 Yacine Ait-Sahalia Sandra Decker John C. Heaton Jong-Wha Lee Andrew Samwick • Rosanne Altshuler Patricia DeVries Judith Hellerstein Thomas Lemieux J. Karl Scholz Patricia M. Anderson Francis X. Diebold 19a1 Hendel Margaret Levenstein John Shea Patrick K Asea John DiNardo James R. Hines, Jr. Steven D. levitt Andrea Shepard Andrew Atkeson Kathryn M. E. Dominguez Hilary W. Hoynes Deborah J. Lucas Nachum Sicherman Orazio Attanasio Steven N. Durlauf Jennifer Hunt Robin 1. Lumsdaine Hilary Sigman Laurence C. Baker Janice C. Eberly Guido Imbens Richard K Lyons Matthew J. Slaughter Richard Baldwin Aaron Edlin Douglas A. Irwin Brigitte Madrian Kathryn E. Spier Susanto Basu Nada Eissa Christine M. Jolls Rodolfo Manuelli Douglas O. Staiger David S. Bates Glenn Ellison Larry E. Jones Kiminori Matsuyama Scott Stem Paul Beaudry Eduardo M. Engel Chinhui Juhn Mark B. McClellan Ann Huff Stevens Geert Bekaert Eric M. Engen Thomas J. Kane· Kathleen M. McGarry Valerie Y. Suslow Roland Benabou William N. Evans Shawn E. Kantor David Meltzer Deborah 1. Swenson Eli Berman Jonathan Feinstein Ani! K Kashyap Gilbert E. Metcalf Alan Taylor Steven T. Berry Joseph P. Ferrie Patrick Kehoe Derek Neal Scott Taylor Giuseppe Bertola Rachel M. Friedberg Daniel Kessler Thomas J. Nechyba Linda Tesar Timothy J. Besley Jordi Gali Sukkoo Kim Steven Olley Aaron Tomell MarkBils David Genesove Nobuhiro Kiyotaki Leslie E. Papke Daniel Trefler Gordon Bodnar William M. Gentry Michael W. Klein Canice Prendergast Robert Triest Lael Brainard· Alec I. Gershberg Michael M. Knetter Thomas J. Prusa Peter Tufano Moshe Buchinsky Robert Gertner Sanders Korenman Daniel Raff Andres Velasco Charles Calomiris Donna B. Gilleskie Samuel S. Kortum Valerie A. Ramey Joel Waldfogel Andrew Caplin Edward 1. Glaeser Carsten Kowalczyk Sergio Rebelo James Walker Eliana A. Cardoso· Sherry A. Glied Michael Kremer Nancy Reichman Jiang Wang Christopher D. Carroll Linda S. Goldberg· Douglas 1. Kruse Paul Rhode Shang-Jin Wei Anne Case Paul Gompers Saul Lach Joshua Rosenbloom David N. Weil Alessandra Casella Lawrence H. Goulder Francine Lafontaine Joshua Rosett Ingrid M. Werner Judith A. Chevalier Shane Greenstein David Laibson Nouriel Roubini Holger C. Wolf lain M. Cockburn Jonathan Gruber Robert J. laLonde Cecilia E. Rouse Alwyn Young Dora Costa Maria J. Hanratty Owen Lamont John P. Rust Stanley E. Zin . Janet Currie Gordon H. Hanson Jenny Lanjouw Xavier Sala-i-Martin Luigi Zingales David M. Cutler Ann Harrison John V. Leahy ·On Leave for Government Service •

30. NBER Reporter Fall 1995 Romer investigates the puzzling cases, to 1273. They fmd that the economic or political crisis. fact that prices rose during the volatility and persistence of devi­ Also participating in this work­ mid- and late 1930s despite the ex­ ations from the law of one price shop were: Robert B. Barsky, tremely depressed level of output have been remarkably stable over NBER and University of Michigan; in those years. She shows that time. Those deviations are correlat­ Olivier J. Blanchard, NBER and throughout the twentieth century, ed highly across commodities (es­ MIT; Mario Crucini, Ohio State Uni­ inflation has depended largely on pecially at annual horizons) and, versity; Robert J. Gordon, NBER the rate of growth of output, and for most pairwise comparisons in and Northwestern University; Rich­ that this growth-rate effect is large­ most centuries, are at least as vola­ ard Grossman, Wesleyan Universi­ ly attributable to the impact of out­ tile as relative prices across differ­ ty; Christopher Hanes, University put growth on materials prices. ent goods within the same country. of Pennsylvania; Michael W. Klein, The inflation during 1934-42 was This analysis challenges the con­ NBER and Tufts University; N. largely a consequence of the rapid ventional view that the modern Gregory Mankiw, NBER and Har­ growth of output during the recov­ floating exchange rate experience vard University; Mancur Olson, ery from the Great Depression, Ro­ is exceptional in terms of the be­ University of Maryland; Stefan Op­ mer finds. She also shows that the havior of relative.prices (adjusted for pers, University of Michigan; Scott National Recovery Administration, exchange rates) across countries. Sumner, Bentley College; Richard by mandating cost-based pricing, Historians have maintained for a Sylla, NBER and New York Uni­ weakened the usual effect that out­ long time that the key to the suc­ versity; and Alan Taylor, North­ put's deviation from trend has on cessful mobilization of resources in western University. prices. the United States during World War Although its role has been over­ II was the Controlled Materials Plan looked by monetary historians, a instituted by the War Production two cent tax on bank checks effec­ Board in July 1943. However, a tive from June 1932 through De­ close look by Rockoff reveals that cember 1934 appears to have been the plan had, at most, a very limit­ an important contributing factor to ed impact on trends in the produc­ that period's severe monetary con­ tion of munitions. This finding sug­ traction. According to estimates by gests that traditional interpretations Lastrapes and Selgin, the tax ac­ of the role of "central planning" counted for between 11 percent during the mobilization may be in and 17 percent of the total increase need of revision. in the ratio of currency to demand Temin writes that a "Korea­ deposits, and for between 11 per­ boom" is thought to have taken cent and 19 percent of the total de­ place in West Germany immediate­ cline in M1 between October 1930 ly following the outbreak of the and March 1933. The contraction­ Korean War on June 26, 1950. ary consequences of the check tax However, the postwar German re­ had been anticipated by many leg­ covery was not beholden to fortu­ islators, but they were unable to itous exogenous shocks, or at least prevent the measure from being in­ not to that particular shock. Neither cluded in the Revenue Act of 1932. German exports nor American im­ Froot, Kim, and Rogoff exam­ ports rose unusually at that time. ine annual data on commodity pri­ Instead, Germany suffered an ad­ ces from England and Holland over verse price shock at the same time a span of seven centuries. Their da­ that its imports increased. The re­ taset incorporates prices for eight sult was not a boom, but a bal­ commodities-barley, butter, cheese, ance-of-payments crisis. The fledg­ eggs, oats, peas, silver, and wheat ling European l'ayments Union act­ ,. -with pound/shilling nominal ex­ ed swiftly to keep this small prob­ y change rates going back, in some lem from escalating into a major

NBER RepotterFalll995 31. •

There is reliable evidence that recently developed techniques in ties; and negative-mean jump mod­ simple rules used by traders have nonparametric analysis, Ait-Saha­ els with time-varying jump frequen­ some predictive value over the fu­ lia and Lo construct an estimator cies. Using data on S&P 500 futures ture movement of foreign ex­ for the SPD implicit in financial as­ options for 1988-93, Bates COnflIffiS change prices. LeBaron reviews set prices, and derive an asymptot­ that shocks to volatility and level some of this evidence and discuss­ ic sampling theory for this estima­ are correlated substantially nega­ es the economic magnitude of this tor to gauge its accuracy. They tively. Still, the stochastic volatility predictability. He then uses inter­ show that the SPD estimator can model explains implicit negative vention data from the Federal Re­ be extracted successfully from op­ skewness of distributions only un­ serve to analyze the profitability of tion prices, and present an empiri­ der extreme parameters that are im­ these trading rules in connection cal application to S&P 500 index plausible given the time-series with central bank activity. His ob­ options for which the SPD estimate properties of implicit spot vari­ jective is to learn to what extent and the corresponding nonpara­ ances. By contrast, the stochastic foreign exchange predictability can metic option pricing formula ex­ volatility/jump diffusion model be confmed to periods of either high hibit volatility "smiles" and other generates substantially more plausi­ or low central bank activity~ His re­ empirical regularities. Options with ble parameter estimates. sults indicate that after removing different times to expiration yield Dechow and Sloan test com­ periods in which the Federal Re­ the family of SPDs over different peting explanations for the higher serve is active, exchange rate pre­ horizons. stock returns generated by "con­ dictability is reduced dramatically. Bates shows that since the 1987 trarian" investment strategies, in­ Implicit in the prices of traded crash, implicit distributions have cluding naive expectations about financial assets are Arrow-Debreu been skewed strongly negatively. future earnings growth and higher state prices or, in the continuous­ He examines two competing expla­ expected returns to compensate for state case, the state-price density nations for this: stochastic volatility risk. Unlike Lakonishok, Shleifer, (SPD) that may be used to price all models with negative correlations and Vishny (1994), they find no assets, traded or nontraded. Using between market levels and volatili- systematic evidence that returns

32. NBER Reporter Fall 1995 stem from naive extrapolation of the asset-pricing kernel: the Abel's tial equilibrium model of portfolio past growth. Rather, consistent with (1990) "Catching Up with the choice in which investors face in­ Laporta (995)' they find that prices Joneses" preferences, and Constan­ come risk that is not directly insur­ naively impound analysts' forecasts tinides's (1990) habit-formation able. They consider the sensitivity of future earnings growth. Howev­ preferences. They find that while of savings and portfolio allocation er, this accounts for only half of the neither model performs well at the rules to different assumptions returns to contrarian investment quarterly horizon, both models about the stochastic process for in­ strategies. The remaining returns perform fairly well when the hori­ come and asset returns, and market can be explained by estimates of zon is lengthened to one or two frictions (transactions costs and expected return generated by a years. In particular, the Hansen­ short sale constraints). Under a cer­ simple valuation model, the au­ Jagannathan restrictions are not vi­ tain assumption, their model pre­ thors show. olated, unconditional first- and sec­ dicts that investors borrow and in­ Daniel and Marshall use co­ ond-moment restrictions are satis­ vest all of their savings in stocks. herence analysis and bandpass fil­ fied, the mean and variance of the This implication holds even with tering analysis to show that, while conditional cov~riance between significant transactions costs in the there is a complete lack of correla­ equity returns an? the pricing ker­ stock market, and a wedge be­ tion between asset returns and nel approximate the corresponding tween the borrowing and lending consumption growth at frequencies moments of the conditional equity rate. When preferences exhibit higher than about 0.7 years (or premium, and variation in the con­ habit persistence, the model pre­ swings longer than 1.4 years), the ditional equity premium is reflected dicts that individuals hold some coherence/correlation between the to some extent in variation in the bonds as well as stocks. However, two series at cyclical frequencies is conditional covariance series. the model with habit persistence above 60 percent. They consider Heaton and Lucas examine the implies a substantial amount of two consumption-based models of quantitative implications of a par- trading in the stock market.

NBER ReponerFall 1995 33. Duffee uses no-arbitrage mod- dure for evaluating any such method and 2) not all the risks that banks els ·to estimate the price of default The evaluations of volatility face can be hedged frictionlessly in risk for 120 firms. With the cross- forecasts most meaningful to users the capital market. This approach sectional (term structure) and time- allows the authors to show how • of forecasts are those conducted series properties of corporate and under economically relevant loss bank-level considerations about Treasury bonds, he can estimate: functions. Lopez proposes a fore- risk management should factor into probabilities of default; the volatili- cast evaluation framework that in- the pricing of those risks that can- ties of these probabilities; their corporates a general class of eco- not be hedged easily. They exam- cross-correlations; and the correla- nomic loss functions. The user's ine several applications, including tions of default probabilities with loss function specifies the three the evaluation of proprietary trad- default-free interest rates. He fmds key elements of the evaluation ing operations, and the pricing of that monthly shocks to investment- framework: the economic events to unhedgeable derivatives positions. grade firms' probabilities of default be forecast; the criterion with Geanakoplos, Zame, and Du- are large, and are correlated posi- which to evaluate those forecasts; bey present a straightforward ex- tively across firms (typical correla- and the subsets of the forecasts of tension of general equilibrium the- tions are about 0.3), However, particular interest. Volatility fore- ory that incorporates default and col- these shocks are correlated only casts are transformed into probabil- lateral, and thereby permits them to minimally with shocks to default- ity forecasts of the specified events, assess analytically the importance free rates. This implies that the as- and the probability forecasts are of these markets to the economy as sumption of independence between evaluated using statistical criteria, a whole. They show that default default probabilities and default- including probability scoring rules, and collateral, although they are free rates is probably appropriate. tailored to the user's interests. Us- suggestive of disequilibrium, pose Crnkovic and Drachman have ing exchange rates, Lopez illus- no special technical problems: un- developed a tool that allows them trates the procedure and confums der the hypotheses on agent be- to evaluate and compare different that the choice of loss function di- havior and foresight that are stan- approaches to market risk mea- recdy affects the forecast evalua- dard in the general equilibrium lit- surement. Their paper describes tion results. erature, collateral and default equi- the ideas behind the tool and illus- Froot and Stein develop a librium always exist. The main the- trate some of its uses. They argue framework for analyzing the deci- sis here is that the scarcity of collat- • that the quality of any market risk sions about capital allocation and eral exerts a powerful influence on measurement system is determined capital structure facing financial in- the economy, inhibiting investment to a large degree by the method stitutions. Their model incorporates and risksharing, and creating incen- used to forecast the probability two key features: 1) value-maximiz- tives to find innovations that econ- density functions of the market va- Ing banks have a well-founded omize on collateraL riables, and they describe a proce- concern with risk management;

e.' ·

34. NBER Reporter Fall 1995 T" I I I t Card and Robins evaluate an immigrant's native culture and lan­ panded trade with countries that earnings subsidy program offered guage are represented broadly in are abundant in less-skilled work­ to long-term welfare recipients in his new country. Further, when ers, as well as with some models of two Canadian provinces. The pro­ governments protect minority inter­ technological change. gram-known as the Self-Sufficien­ ests dIrectly, incentives to be as­ cy Project-provides a subsidy equal similated into the majority culture to half the difference between an are reduced. Both factors may ex­ earnings target ($2500 or $3083 per plain the recent rise in multicultur­ month, depending on the province) alism. Lazear tests this theory and Additional Papers and the individual's actual earnings. confirms it by examining U.S. Cen­ It is similar to a negative income sus data. He shows that the likeli­ tax with two important differences: hood that an immigrant will learn Additional Papers are not official first, eligibility is limited to former English is related inversely to the NBER Working Papers but are list­ welfare recipients who find a full­ proportion of the local population ed here as a convenience to NBER time job (30 hours per week or that speaks the immigrant's native researchers and prospective read­ more); second, the subsidy pay­ language. ers. Additional Papers are free of ment is based on individual earn­ A structural change appears to charge to Corporate Associates. For ings rather than family income. For have occurred in U.S. collective all others there is a charge of a group of 2000 individuals ob­ bargaining in the 1980s. Tracy and $5.00 per Additional Paper re­ quested. (Outside of the United served over 18 months, early find­ Cramton ask how the hiring of re­ ings suggest that the financial in­ placement workers explains this States, add $10.00 for postage and handling.) Advance pay­ centives of the Self-Sufficiency Pro­ change. For a sample of over 300 ment is required on all orders. gram significantly increase labor major strikes since 1980, they esti­ please do not send cash. Request market attachment and significantly mate the likelihood of replace­ Additional Papers by name, in writ­ reduce welfare participation. ments being hired. They find that ing, from: Additional Papers, NBER, If profit-maximizing firms have the risk of being replaced declines t 1050 Massachusetts Avenue, Cam­ limited information about the gen­ during tight labor markets, and is eral productivity of new workers, lower for bargaining units with bridge, MA 02138-5398. they may choose to use easily ob­ more experienced workers. Fur­ "Reports of the Dollar's Death Are servable characteristics, such as ther, the use of strikes decreases Greatly Exaggerated," by Jeffrey A. years of education or ethnic back­ significantly as the predicted risk of Frankel. This paper is forthcom­ ground, to "statistically discrimi­ replacement increases. Tracy and ing, under the title, "Still the Lingua nate" among workers. As firms ac­ Cramton estimate that a ban on the Franca: The Exaggerated Death of quire more information about a use of replacement workers would the Dollar," in Foreign Affairs, 74, worker, they may become more have increased the incidence of no. 4, July/August 1995. dependent on actual productivity strikes from 1982-9 by 3 percent­ "European Integration and the Re­ and less dependent on easily ob­ age points, a 30 percent increase. gionalization of World Trade and servable characteristics or creden­ Krueger examines the relation­ Currencies: The Economics and the tials that predict productivity. AI­ ship between price growth and Politics," by Jeffrey A. Frankel tonji and Pierret formalize these skill intensity across 150 manufac­ and Shang-Jin Wei This paper is ideas and examine some of their turing industries between 1989 and forthcoming in Monetary and Fis­ empirical implications. They then 1995. He finds that wage growth cal Policy in an Integrated Europe, show that even if employers learn and increases in intermediate edited by Barry Eichengreen, Jeffry about the productivity of new work­ goods prices are passed through to Frieden, and Jurgen von Hagen, ers relatively slowly, the portion of final product prices roughly in pro­ from Springer-Verlag Press, New the return to education that could portion to their factor shares. Sec­ York and Heidelberg, 1995. reflect signaling of ability is limited. ond, product prices have grown "Foreign Direct Investment and Common culture and language relatively less in sectors that use Politics: The Swedish Model," by facilitate trade and interaction less-skilled labor more intensively. Magnus BlomstrOm and Ari Kokko. .. among individuals. However, as­ That finding is consistent with the .• similation is less likely when an Stolper-Samuelson theory of ex-

NBER Reporter Fall 1995 35. \"

Reprints Available and Alan B. Krueger (NBER Tech- seph E. Stiglitz (NBER Working nical Paper No. 150) Paper No. 3289) The following NBER Repri'nts, 1973. "Did Teachers' Verbal Ability 1984. The Dynamic-Optimizing Ap- intended for nonprofit education and Race Matter in the 1960s? Cole- proach to the Current Account: and research purposes, are now man Revisited," by Ronald F. Ehr- Theory and Evidence," by Assaf available. (Previous issues of the enberg and Dominic J. Brewer Razin (NBER Working Paper No. NBER Reporter list titles 1-1967 and (NBER Working Paper No. 4293) 4334) contain abstracts of the Working 1974. "Resisting Migration: Wage 1985. "Pensions and Retirement: Papers cited below.) Rigidity and Income Distribution," Evidence from Union Army Veter- These reprints are free of charge by Assaf Razin and Efraim Sadka ans," by Dora L. Costa (NBER to Corporate Associates. For all (NBER Working Paper No. 4903) Working Paper No. 4537) others there is a charge of $5.00 1975. "Labor Demand and the 1986. "Is a Yen Bloc Emerging?" by per reprint requested. (Outside Source of Adjustment Costs," by Jeffrey A. Frankel and Shang-Jin of the United States, add $10.00 Daniel S. Hamermesh (NBER Wei (NBER Working Paper No. for postage and handling.) Ad­ Working Paper No. 4394) 4050) vance payment is required on 1976. "Tax Projections and the 1987. "Policies to Encourage In- all orders. please do not send Budget: Lessons from the 1980s," flows of Technology Through For- cash. Reprints must be requested by by AlanJ. Auerbach (NBER Work- eign Multinationals," by Ari o. number, in writing, from: Reprints, ing Paper No. 5009) Kokko and Magnus Blomstrom NBER, 1050 Massachusetts Avenue, 1977. "Quality-Adjusted Cost Func- (NBER Working Paper No. 4289) Cambridge, MA 02138-5398. tions for Child Care Centers," by H. 1988. "State Responses to Fiscal ~ Naci Mocan (NBER Working Pa- Crises: The Effects of Budgetary In- 1968. "Do Teachers' Race, Gender, per No. 5040) stitutions and Politics," by James and Ethnicity Matter? Evidence from the National Educational Longitudi- 1978. "Two-Stage Least Squares Es- M. Poterba (NBER Working Paper nal Study of 1988," by Ronald G. timation of Average Causal Effects No. 4375) Ehrenberg, Daniel D. Goldha- in Models with Variable Treatment 1989. "The Decline of Traditional ber, and Dominic J. Brewer Intensity," by Joshua D. Angrist Banking: Implications for Financial (NBER Working Paper No. 4669) and Guido W. Imbens (NBER Stability and Regulatory Policy," by 1969. "Precautionary Saving and Technical Paper No. 127) Franklin R. Edwards and Frede- Social Insurance," by R. Glenn 1979. "Two Fallacies Concerning ric S. Mishkin (NBER Working Pa- Hubbard, Jonathan S. Skinner, Central Bank Independence," by per No. 4993) and Stephen P. Zeldes (NBER Bennett T. McCallum (NBER 1990. "Debt and Seniority: An Anal- Working Paper No. 4884) Working Paper No. 5075) ysis of the Role of Hard Claims in 1970. "Following in Her Footsteps? 1980. "A Retrospective on the Debt Constraining Management," by Oli- Faculty Gender Composition and Crisis," by Michael P. Dooley ver Hart and John Moore (NBER Women's Choices of College Ma- (NBER Working Paper No. 4963) Working Paper No. 4886) jors," by Brandice J. Canes and 1981. "A Model of the Optimal 1991. "Energy Tax Credits and Res- Harvey S. Rosen (NBER Working Complexity of Legal Rules," by idential Conservation Investment: Paper No. 4874) Louis Kaplow (NBER Working Pa- Evidence from Panel Data," by 1971. "Tastes and Technology in a per No. 3958) Kevin A. Hassett and Gilbert E. Two-Country Model of the Busi- 1982. "A Sticky-Price Manifesto," by Metcalf (NBER Working Paper No. ness Cycle: Explaining International Laurence M. Ball and N. Gregory 4020) Comovements," by Alan C. Stock- Mankiw (NBER Working Paper 1992. "Information and Economic man and Linda L. Tesar (NBER No. 4677) Efficiency," by Richard J. Arnott, Working Paper No. 3566) 1983. "Randomization in Optimal Bruce C. Greenwald, and Joseph 1972. "Split-Sample Instrumental Income Tax Schedules," by Dago- E. Stiglitz (NBER Working Paper Variables Estimates of the Return to bert L Brito, Jonathan H. Hamil- No. 4533) g l...-S_ch_O_O_I_in_g_,'_' _b_y_J_O_s_h_ua_D_._An_grt_·s_t__ t_o_n_,_S_t_e_v_e_n_M_._S _1_u_ts_ky_,_a_n_d_J_o_-__C_R_ep_rt_·_n_ts_c_o_n_ti_·n_u_ ed_o_n_p_a__e_3_Z_9_...J.

36. NBER RejJorterFall 1995 t Reprints, continued Michael D. Bordo, Ehsan U. Choudhri, and AnnaJ. Schwartz Bureau Books 1993. "Participation in and Contri- (NBER Working Paper No. 4481) butions to 401 (k) Pension Plans: Evidence from Plan Data," by Les- 2004. "Internal Finance and Firm lie E. Papke (NBER Working Paper Investment," by R. Glenn Hub- No. 4199) bard, Anil K Kashyap, and Toni M. Whited (NBER Working Paper 1994. "Home-Country Effects of No. 4392) Foreign Direct Investment: Swe- NBER den," by Magnus Blomstrom and 2005. "Executive Pay and Perfor- Macroeconomics Ari o. Kokko (NBER Working Pa- mance: Evidence from the U.S. per No. 4639) Banking Industry," by R. Glenn Annual 1995 Hubbard and Darius palla (NBER 1995. "Econometric Estimates of Working Paper No. 4704) Price Indexes for Personal Comput- ers in the 1990s," by Ernst R. 2006. "Lost Jobs," by Robert E. Hall The NBER Macroeconomics Berndt, Zvi Griliches, and Neal J. 2007. "Expected and Predicted Re- Annual 1995, edited by Ben S. Rappaport (NBER Working Paper alignments: The FP/DM Exchange Bernanke and Julio J. Rotemberg, is No. 4549) Rate During the EMS, 1979--93," by available from the MIT Press for 1996. "Do Doctoral Students' Finan- Andrew K. Rose and Lars E. o. $20.00 in paperback or $40.00 in cial Support Patterns Affect Their Svensson (NBER Working Paper clothbound version. This is the Times-to-Degree and Completion No. 3685) tenth in a series of annuals de­ Probabilities?" by Ronald G. Ehr- 2008. "General Purpose Technolo- signed to stimulate research on enberg and panagiotis G. Mavros gies: 'Engines of Growth?'" by problems in applied macroeco­ (NBER Working Paper No. 4070) Timothy F. Bresnahan and Man- nomics, to bring frontier theoretical 1997. "A Semi-Classical Model of uel Trajtenberg (NBER Working developments to a wider audience, Price Level Adjustment," by Ben- Paper No. 4148) and to accelerate the interaction nett T. McCallum (NBER Working 2009. "Interest Allocation Rules, Fi- between analytical and empirical Paper No. 4706) nancing Patterns, and the Opera- research in macroeconomics. -1998. "International Public Opinion tions of U.S. Multinationals," by Among the topics discussed in on the Environment," by David E. Kenneth A. Froot and James R. this volume are: wage inequality Bloom Hines, Jr. (NBER Working Paper and regional unemployment; capi­ 1999. "The Growth of Nations," by No. 4924) tal utilization and returns to scale; N. Gregory Mankiw 2010. Taxes, Technology Transfer, banks and derivatives; exchange­ 2000. "Training, Wage Growth, and and the Rand D Activities of Multi- rate-based stabilizations; and infla­ Job Performance: Evidence from a national Firms," by James R. tion indicators and policy. Company Database," by Ann P. Hines, Jr. (NBER Working Paper Bernanke and Rotemberg are Bartel (NBER Working Paper No. No. 4932) both research associates in the 4027) 2011. "Incentives, Optimality, and NBER's economic fluctuations pro­ 2001. "Unemployment Benefits and Publicly Provided Goods: The Case gram. Bernanke is also a professor Labor Market Transitions: A Multi- of Mental Health Services," by at the Woodrow Wilson School of nomial Logit Model with Errors in Richard G. Frank and Martin Public and International Mfairs at Classification," by James M. Poter- Gaynor (NBER Working Paper Princeton University. Rotemberg is ba and Lawrence H. Summers No. 3700) a professor at the Sloan School of (NBER Working Paper No. 4434) 2012. "Efficient and Inefficient Management, MIT, and Harvard 2002. "Economic Reform and the Sales of Corporate Control," by Lu- Business School. Process of Global Integration," by cian Arye Bebchuk (NBER Work- Jeffrey D. Sachs and Andrew M. ing Paper No. 4788) Warner (NBER Working Paper No. 2013. "Is There a 'Credit Channel' 5039) for Monetary Policy?" by R. Glenn 2003. "Could Stable Money Have Hubbard (NBER Working Paper ) Averted the Great Contraction?" by No. 4977)

NBER Reporter Fall 1995 37. •

The Industrial Organization and Regulation of the Securities Industry

The Industrial Organization and Regulation oftbe Securities Industry, edited by Andrew W. Lo, now may be ordered from the University of Chicago Press for $59.95. In recent years, regulation Individual has played a crucial role in a num­ and Social ber of industries, including the air­ lines, telecommunications, trans­ Works Councils • Responsibility portation, and public utilities. However, despite a plethora of studies of and headlines about the Works Councils: Consulta­ Individual and Social Re­ securities industry, surprisingly lit­ tle has been written about the tion, Representation, and Coop­ sponsibility: Cbild Care, Educa­ edited by Joel Rogers and industrial organization and regula­ eration, tion, Medical Care, and Long­ Wolfgang Streeck, is a comprehen­ tion of that industry. This volume Term Care in America, edited by sive study of various employee represents a step in that direction Victor R. Fuchs, will be available representation and consultation and should interest both econo~ from the University of Chicago agreements in Europe and North mists and participants in the securi­ Press this fall. This volume is the America. The countries studied in­ ties markets. result of an NBER conference in clude Germany, the Netherlands late 1994 that brought together ap­ Lo is a research associate in the France, Spain, Sweden Italy poland' plied and theoretical economists NBER's Program in Asset Pricing Canada, and the Unit~d St;tes. Thi~ lawyers, physicians, and SDcial sci~ and the Harris & Harris Group Pro­ volume is part of the NBER Com­ entists to discuss issues in the pro­ fessor of Finance at MIT's Sloan parative Labor Markets Series. School of Management. vision of human services. Priced at Works Councils is available now $55.00, the book should be of inter­ from the University of Chicago est to a broad spectrum of readers Press, and priced at $55.00. Both and not just to academic economists.' authors are at the University of Fuchs is a research associate in :Visconsin, Madison, where Rogers and former director of, the NBER'~ 1S professor of law, political sci­ Program in Health Care. He is also ence, and sociology, and Streeck is Henry J. Kaiser, Jr. Professor Emeri­ professor of sociology and industri- tus at Stanford University. al relations. eJ

38. NBER Reporter Fall 1995 cess to college. Using several sour­ . Current Working Papers ces of variation in costs, this paper evaluates the price sensitivity of ------youth regarding schooling. Most of the evidence points to large im­ pacts on enrollment, particularly for low-income students and for those attending two-year colleges. The states have chosen to pro­ mote enrollment in college by keeping tuition low through across­ the-board subsidies rather than by using more targeted, means-tested aid. As public enrollments increase, this becomes an expensive strategy. However, the evidence on enroll­ ment responses to means-tested aid is much weaker than the response to broad subsidies. After a federal means-tested grant program was established in 1973, for example, there was no disproportionate in­ crease in college enrollment by low-income )fouth.

Leverage, Investment, and Firm Growth Larry Lang, Eli Ofek, and Rene M. Stulz NBER Working Paper No. 5165 July 1995 Asset Pricing, Corporate Finance We show that there is a negative relationship between leverage and future growth at the fum level, and for diversified finns at the segment level. Further, this negative rela­ tionship holds for firms with low Tobin's q, but not for finns with a high q, or in high-q industries. NBER Working Papers Labor Studies Therefore, leverage does not re­ Although economists have spent duce growth for firms known to Rising Public College the past decade analyzing the ris­ have good invesqnent opportuni­ Tuition and College Entry: ing payoff to schooling, they know ties, but it is negatively related to How Well Do Public much less about how youth re­ growth for firms whose opportuni­ Subsidies Promote sponds to that payoff, or how ef­ ties either are not recognized by Access to College? fective policies aimed at influenc­ the capital markets or are not suffi­ Thomas]. Kane ing those decisions are. Yet state ciently valuable to overcome the NBER Working Paper No. 5164 and federal governments currently effects of their debt overhang. ,~ July 1995 spend more than $53 billion annu­ , JEL Nos. 120, 128 ally, hoping to promote greater ac-

NBER ReporterFall 1995 39. Why Is There a Home Bias? a large number of diversilled inves­ level of trade volume give rise to • . An Analysis of Foreign tors taking small positions against protection, implying that whether Portfolio Equity its mispricing. In reality, however, rising imports are met with greater Ownership inJapan arbitrage is conducted by a rela­ liberalization or increased protec- Jun-Koo Kang and tively small number of highly spe­ tion depends on whether they are Rene M. stutz cialized investors who take large part of a cyclical upward trend in NBER Working Paper No. 5166 positions using other people's trade volume or an acyclical in- July 1995 money. Such professional arbitrage crease in import levels. Asset Pricing, Corporate Finance has a number of interesting impli­ cations for security pricing, includ­ Worker Adaptation and We use data on foreign stock ing the possibility that it becomes Employer Accommodation ownership in Japan from 1975 to ineffective in extreme circum­ Following the Onset of 1991 to examine the determinants stances, when prices diverge far a Health Impairment of the home bias in portfolio hold­ from fundamental values. The Mary C. Daly andJohn Bound ings. Existing models of interna­ model also suggests where anom­ NBER Working Paper No. 5169 tional portfolio choice that predict alies in financial markets are likely July 1995 that foreign investors hold national to appear, and why arbitrage fails JEL Nos. J14, J26 market portfolios, or portfolios tilt­ to eliminate them. Aging, Labor Studies ed toward stocks with high expect­ ed returns, are inconsistent with Protection and the Using data from the new Health the evidence we provide. We show Business Cycle and Retirement Survey, we investi­ that foreign investors place too Kyle Bagwell and gate the responses of workers and much weight on shares of fIrms in Robert W. Staiger their employers to the onset of work-limiting health impairments. manufacturing industries, large NBER Working Paper No. 5168 We find that many workers who fIrms, fums with good accounting July 1995 suffer from health limitations are performance, ftrms with low unsys­ JEL No. F13 tematic risk, and firms with low International Trade and Investment accommodated directly by their leverage. Controlling for size, it ap­ employers: those who do not re­ pears that small fIrms that export Empirical studies repeatedly ceive direct accommodation fre­ • more have greater foreign owner­ have documented the countercycli­ quently adapt to their limitations ship. Foreign investors do not per­ cal nature of trade barriers. In this by altering their job demands or by form significantly worse than if paper, we propose a simple theo­ changing jobs. These ftndings point they held the Japanese market retical framework consistent with to the potential for adjustments on portfolio, however. Mter control­ this and other empirical regularities both sides of the market: by em­ ling for ftrm size, we fInd no evi­ in the relationship between protec­ ployers, in the form of job accom­ dence that foreign ownership is re­ tion and the business cycle. We ex­ modation, and by employees in the lated to expected returns of shares. amine how countries can maintain form of job change. We show that a model with size­ efficiency-enhancing reciprocal based informational asymmetries trade agreements that control their and deadweight costs can yield as­ temptation to resort to beggar-thy­ Industry or Class set allocations consistent with our neighbor poliCies, requiring that Cleavages Over Trade evidence. such' agreements be self-enforcing. Policy? EvideQ.ce from We find theoretical support for the British General The Limits of Arbitrage countercyclical movements in lev­ Election of 1923 els of protection as the rapid Andrei Shieifer and Douglas A. Irwin growth in trade volume that is as­ Robert W. VlShny NBER Working Paper No. 5170 sociated with a boom phase facili­ NBER Working Paper No. 5167 July 1995 tates the maintenance of more lib­ July 1995 International Trade and Investment eral trade policies that are sustain­ Asset Pricing able during a recession phase with Economists and political scien- In traditional models, arbitrage slow growth. However, we also tists frequently have attempted to ." in a given security is performed by fInd that acyclical increases in the determine whether political action

40. NBER ReponerFalll995 .~ related to trade policy takes place the passive loss limitations was The Seasonality of along factor lines (such as capital secondary to that of other reforms Consumer Prices versus labor, as implied by the Stol­ enacted by TRA, most importantly Michael F. Bryan and per-Samuelson theorem) or along the repeal of the investment tax Stephen G. Cecchetti industry lines (as implied by mod­ credit and the exclusion of long­ NBER Working Paper No. 5173 els with imperfect factor mobility). term capital gains. These other re­ July 1995 This paper examines voting pat­ forms not only lowered aftertax JEL Nos. E31, E32 terns in the British general election rates of return on tax sheltered in­ Monetary Economics of 1923, which hinged on free vestments, but also eliminated the trade versus protection. This elec­ positive correlation between the in­ In this paper, we reevaluate the tion provides an opportunity to dis­ vestor's marginal tax rate and the evidence of seasonality in prices tinguish between the two hypothe­ investment's aftertax rate of return. which we find to be substantially ses, because either an industry or a As a result, high income taxpayers greater than previous research has factor alignment among voters is ceased to be the natural clientele indicated. That is, seasonal price plausible: rigidities in the interwar for legitimate tax shelters after movements have become more labor market often have been dis­ TRA. The passi~e loss rules were prominent in the relatively stable cussed, and electoral politics often more effective i;). curtailing the use inflation environment that has pre­ has been viewed as having a class of "abusive" tax shelters; however, vailed since 1982. basis. I exploit data from the British a more narrowly focused restriction We draw one main conclusion census of 1921, which divides the on seller financing of tax sheltered from this analysis: the amount of population into categories of occu­ investments could have accom­ seasonality in prices differs greatly pation (by industry) and economic plished the same goal with much by item, making it difficult to gen­ class (by income and/or skillleveO, less scope for discouraging produc­ eralize about seasonal price move­ and relate these data to cross-coun­ tive economic investments. ments. A casual reading fails to re­ ty variation in voting for the con- veal one easily identifiable origin •• tending political parties. My results The Term Structure of of the seasonal variation of prices . • indicate that county differences in Interest Rates in a Pure That is, seasonality in consumer the occupational structure of the Exchange Economy with prices is predominantly idiosyncrat­ electorate account for the election Heterogeneous Investors ic in nature, a result that contrasts outcome better than differences in Jiang Wang with studies demonstrating a com­ class structure. NBER Working Paper No. 5172 mon seasonal cycle in real eco­ July 1995 nomic variables. Tax Shelters and Passive Asset Pricing This finding has an important Losses After the Tax practical implication: given the se­ .Reform Act of 1986 This paper presents an eqUilibri­ um model of the term structure of lective, disaggregated approach Andrew A. Samwick interest rates when investors have taken by the Bureau of Labor Sta­ NBER Working Paper No. 5171 heterogeneous preferences. The tistics to adjust data seasonally, the July 1995 basic model considers a pure ex­ existence of idiosyncratic seasonali­ JEL Nos. H24, E62 change economy of two classes of ty increases the likelihood of al­ Public Economics investors with different· (but con­ lowing noise in the aggregate CPI The precipitous decline in tax stant) relative risk aversion, and at a seasonal frequency. This ar­ sheltered investments after the Tax gives closed-form solutions to bond gues in favor of seasonally adjust­ Reform Act of 1986 (TRA) is attrib­ prices. I use the model to examine ing the index afte~ aggregation. uted widely to the passive loss the effect of heterogeneity of pref­ rules. These rules disallowed losses erence in the behavior of bond Is an Integrated Regional from activities in which the taxpay­ yields. I also consider extensions to Labor Market Emerging in er did not participate materially as cases of more than two investors. East and Southeast Asia? a current deduction against all David E. Bloom and sources of income except for other WaseemNoor passive activities. This paper dem- NBER Working Paper No. 5174 _ onstrates instead that the role of July 1995

NBER ReponerFalll995 41. International Trade and Investment, nontraded goods in the utility func­ In Search of Empirical Labor Studies tion. Further, our analysis uncovers Evidence That Links a second puzzle: the composition We examine labor market inte­ Rent and User Cost of investors' portfolios appears to gration in east and southeast Asia Dixie M. Blackley and be strongly at variance with the James R. FolIain (ESEA) during the 1980s, focusing predictions of the model that incor­ on intraregional labor mobility and NBER Working Paper No. 5177 porates nontraded goods. the two other main channels of in­ July 1995 tegration: capital mobility and Most models of the rental hous­ trade. We find that labor market in­ Should the Stagnant ing market assume a close linkage tegration increased sharply among lIonrreo~ershipRate between- the level of residential ESEA countries in the 1980s, with 9 Be a Source of Concern? rents and the aftertax user cost of percent of ESEA's labor force par­ Richard K. Green rental housing capital. However, ticipating in cross-national labor NBER Working Paper No. 5176 little empirical evidence exists to market transactions in 1991, either July 1995 establish the strength of this link­ directly via labor mobility or indi­ JEL No. R21 age, or the speed with which rents rectly via capital mobility or trade, The homeownership rate in the adjust to changes in user cost or up from just 5.2 percent in 1980. United States was essentially stag­ tax policy. In order to shed light We also find that trade is the domi­ nant during the 1980s. This stagna­ on both of these issues, this paper nant mechanism through which re­ tion should be a source of concern develops and estimates an econo­ gional labor market integration oc­ if it reflects stagnant economic con­ metric model of the rental housing curred in the 1980s, with labor mi­ ditions and ownership opportuni­ market. We use U.S. annual data gration contributing only modestly ties, not if it simply reflects chang­ for 1964 through 1993 to generate to the process. ing demographic conditions or two-stage least squares estimates of preferences. a four-equation structural model. Nontraded Goods, Using a series of affordability Although our results are gener- •. Nontraded Factors, measures, I fmd that homeowner­ ally consistent with expectations . and International ship opportunities improved almost and reveal several interesting rela­ Nondiversification everywhere during the 1980s, sug­ tionships among the system vari­ Marianne Baxter, gesting that the cause of the stag­ ables, we fail to identify a strong UrbanJ.Jer.auurun,and nation rate was something other relationShip between rent and user Robert G. King than economic conditions. In fact, I cost. About half of an increase in NBER Working Paper No. 5175 find that both demographics and user cost ultimately is passed along July 1995 changes in preferences led to an as higher rent. The adjustment pro- JEL No. F21 increase in the proportion of cess also takes a long time, with International Finance and households headed by single peo­ only about a third of the long-run Macroeconomics ple; all else being equal, this would effect realized within 10 years of a Can the presence of nontraded tend to push the owner-occupancy user cost shock. The fundamental consumption goods explain the rate downward. I also find that reason for this result is that our es­ high degree of "home bias" in in­ while homeownership opportuni­ timate of the user cost series, based vestor portfolios? No, so long as in­ ties improved during the 1980s, the upon widely accepted procedures, dividuals have access to" free inter­ ex ante use cost of owning a home is much more volatile than the resi­ national trade in financial assets, increased almost everywhere, re­ dential rent series. we find. In particular, it is never ducing the fmancial attractiveness optimal to exhibit home bias with of owning a home. The combina­ Risk-Based Capital respect to domestic traded-good tion of improving affordability and Standards and the equities. By contrast, an optimal worsening financial appeal had an Riskiness of Bank portfolio may exhibit substantial overall neutral effect on the aggre­ Portfolios: Credit home bias with respect to nontrad­ gate ownership rate. and Factor Risks ed-good equities, although this re­ Steven R. Grenadier and sult requires a very low degree of BrianJ.Hall substitution between traded and NBER Working Paper No. 5178 • 42. NBER Reporter Fall 1995 -, July 1995 Expectations, Efficiency, characteristics and collateral (house) Bank risk-based capital (RBC) and Euphoria in the value. Second, we empirically rec­ ognize important interactions be­ standards require banks to hold dif­ Housing Market tween the status of the prepayment ferent amounts of capital for differ­ Dennis R. Capozza and option and the influence of income ent classes of assets, based almost PaulJ.Seguin and collateral constraints on pre­ entirely on credit risk. This paper NBER Working Paper No. 5179 payment behavior. Third, we use a provides both a theoretical and an July 1995 JEL Nos. R21, G14 major source of data that has not empirical framework for evaluating been used previously in examining such standards. We present a mod­ We study expectations of capital the prepayment anonialy: the Amer­ el outlining a pridng methodology appredation in the housing mar­ ican Housing Survey. for loans subject to default risk. ket. Expectations impounded in Among our fmdings are: when The model shows that the returns the rent/price ratio at the begin­ the household is either collateral on such loans are affected by the ning of the decade successfully constrained or income constrained, complicated interaction among the predict appredation rates, we fmd, or the option is likely to be "out of likelihood of default, the conse­ but only if we control first for the money," the influence of the quences of default, term structure cross-sectional "differences in the option value on prepayment be­ variables, and the pridng of factor quality of rental versus owner-oc­ havior is half as large. When the risks in the economy. When we ex­ cupied housing. We also demon­ status of the option and the influ­ amine whether the weights on risk strate that observed rent/price ra­ ence of potential household con­ accurately reflect the risk of bank tios contain a disequilibrium com­ straints are recognized more appro­ assets, we find that they fail even ponent that has the power to fore­ priately, they account for nearly all in their limited goal of correctly cast subsequent rates of appreda­ explanatory power otherwise attrib­ quantifying credit risk. For exam­ tion. Finally, we provide evidence utable to household demographic ple, our findings indicate that the consistent with euphoria: par tid­ characteristics. ..~.. RBC weights overpenalize home pants in housing markets appear to , mortgages, which have an average overreact to income growth. Residential Mobility credit loss of 13 basis points, rela­ and Mortgages tive to commerdal and consumer The Effect of Income and SewinChan loans. The RBC rules also contain a Collateral Restraints NBER Working Paper No. 5181 significant bias against direct mort­ on Residential Mortgage July 1995 gages relative to mortgage-backed Terminations JEL Nos. J60, G 12 securities. In addition, we find Wayne Archer, David C. Ling, large differences in the credit riski­ Mortgage applications are a de­ and Gary A. McGill ness of loans within the 100 per­ tailed and accurate source of house­ NBER Working Paper No. 5180 hold information that is verified by cent weight class, and potentially July 1995 underwriters, making them a more large benefits to loan diversifica­ JEL No. G12 accurate data source than self-re­ tion, neither of which is considered ported survey responses. This pa­ in the RBC regulations. We also ex­ The prepayment behavior of per discusses how mortgage data amine other types of bank risk by home mortgage borrowers is incon­ can be applied to areas of econom­ estimating a simple factor model sistent with behavior implied by ics outside of mortgage finance. As that decomposes loan risk into classical option theory. In attempt­ a supplement to variables from the term structure, default, and market ing to explain the anomaly, a sub­ application form, the self-selection risk. One implication of our find­ stantial literature has emerged ex­ of mortgage po~nts can infer ex­ ings is that although banks have re­ amining the problem, focusing on pected mobility. I estimate a dura­ allocated their portfolios in ways the characteristics of the mortgage tion model of housing spells, and intended by the RBC standards, and on the historic path of interest the points indicator is highly signif­ they may merely have substituted rates. We contribute to that litera­ icant in predicting mobility for low one type of risk (term structure ture in three respects. First, we ex­ loan-to-value borrowers. These risk) for others (default and market plore the influence of household fmdings demonstrate the potential risk), of which the net effect is level characteristics on prepayment fruitfulness of using this new data »unknown. behavior, using both householder source.

NBER ReponerFall 1995 43. Housing Price cal link between appreciation in The results indicate the sensitivi- • Dynamics Within a house prices and the savings be­ ty of default to the initial loan-to- . Metropolitan Area havior of homeowners during the value ratio of the loan and the Karl E. Case and 1980s. I use household asset and course of housing equity. The latter Christopher J. Mayer debt data taken from the 1984 and is a measure of the extent to which NBER Worlcing Paper No. 5182 1989 waves of the Panel Study of the default option is "in the mon- July 1995 Income Dynamics for a sample of ey." The results also indicate the JEL Nos. R12, R21 home owning households with importance of trigger events, name- members under age 65 to construct ly unemployment and divorce, in This paper analyzes the pattern changes in real household wealth affecting prepayment and default of cross-sectional house price. ap­ as a measure of household saving behavior. preciation in the Boston metropoli­ behavior. I use variation in housing tan area from 1982 to 1994. The em­ We use the empirical results to market conditions across time and pirical results are consistent with analyze the costs of a current poli- regions to identify savings effects. many of the predictions of a stan­ cy proposal: stimulating homeown­ dard urban model in which towns My analysis suggests that the es­ ership by offering low-downpay- have a fIxed set of locational attri­ timated marginal propensity to ment loans. We simulate default butes and amenities. In particular, consume out of real housing capi­ probabilities and costs on zero­ the evidence suggests that house tal gains is 0.03 for the median downpayment loans, and compare prices in towns with a large share household saver. However, there is them to conventional loans with of residents working in the manu­ an asymmetry in the saving re­ conventional underwriting stan- facturing sector in 1980 grew less sponse to both total and unantici­ dards. Our results indicate that if quickly in the ensuing years when pated real capital gains on housing. zero-downpayment loans were aggregate manufacturing employ­ All of the savings offset comes priced as if they were mortgages ment fell. As baby boomers moved from households that experience with 10 percent downpayments, into middle age, house values ap­ real capital losses on housing. then the additional program costs preciated faster in towns with a Households that experience real would be 2 to 4 percent of funds ." larger initial percentage of middle­ gains do not change their saving made available, when housing pri- aged residents. Housing values beha vior. The existence of this ces increase steadily. If housing pri- rose more slowly in towns that asymmetry casts doubt on the ces remain constant, the costs of allowed additional construction, power of changes in house prices the program would be much larger and values rose faster in towns to explain the time-series path of indeed. Our estimates suggest that closer to Boston. Finally, as fewer saving in the United States. additional program costs could be families had children who attended between $74,000 and $87,000 per public schools statewide, the price Mortgage Default and Low million dollars of lending. If the ex­ premium associated with housing Downpayment Loans: The pected losses from such a program in towns with good schools fell. All Costs of Public Subsidy were not priced at all, the losses of these fmdings support the view Yongheng Deng, from default alone could exceed 10 that town amenities and public ser­ John M. Quigley, and percent of the funds made avail- vices are not easily replicated or Robert Van Order able for loans. quickly adaptable to shifts in de­ NBER Working Paper No. 5184 mand, even within a metropolitan July 1~95 The Impact of Immigration area. JEL Nos. H29, G12 on American Labor Markets Prior to House Prices and This paper presents a unified the Quotas Homeowner Saving model of the default and prepay­ Timothy J. Hatton and Behavior ment behavior of homeowners in a Jeffrey G. Wil1iamson proportional hazard framework. Gary V. Engelhardt NBER Worlcing Paper No. 5185 The model uses the option-based NBER Worlcing Paper No. 5183 July 1995 approach to analyze default and July 1995 Development of the American prepa yment, and considers these JEL Nos. E21, R31 Economy, Labor Studies two interdependent hazards as This paper examines the empiri- competing risks. Current debate on the impact •

44. NBER Repotter Fall 1995 ) and assimilation of immigrants into Research (ICPSR) transferred the experimental work on auctions: .. the American labor market sounds data to magnetic tape. new entrants alter strategic behav­ remarkably like the debate that A number of researchers have ior mainly on the market's demand eventually triggered the imposition used the ICPSR tape, but two key side, without comparable implica­ of the quotas in the 1920s. Then as problems discourage many from tionsfor the supply side. We also now, observers failed to agree on taking advantage of this unique show that bids and offers are influ­ exactly what the impact of the dataset: First, modem econometric enced by fundamentals and specif­ mass migration was on labor mar­ software does not have the ability ic policy measures. kets. Despite its relevance to cur­ to read the obsoiete ICPSR format rent discussion, there has been al­ Second, the process of transferring Career and Family: College most no quantitative effort to as­ the data from the NBER's handwrit­ Women Look to the Past sess late-19th-century impact, while ten sheets to the ICPSR tape intro­ Claudia Goldin analysis instead has been obsessed duced a number of mistakes. We NBER Working Paper No. 5188 with assimilation issues. This paper have eliminated these two impedi­ July 1995 redresses this imbalance by con­ ments to the us~ of the NBER data­ JEL Nos. Jl, J2 fronting three questions that are set by convertin£ the ICPSR tape to Development of the American just as relevant today as they were a portable format, and by verifying Economy, Labor Studies almost a century ago: Did late- the accuracy of the data using the 19th-century American immigrants NBER's original handwritten sheets. Recent college graduate women act as a flexible (guestworker) la­ The dataset is now available on the express frustration regarding the bor supply? Did they flow into oc­ Internet, and can be accessed using obstacles they will face in combin­ cupations in which job creation standard gopher or web-browser ing career and family. Tracing the was fast, or did they displace na­ software. demographic and labor force expe­ tives in occupations in which job riences of four cohorts of college creation was slow? Did immigrants Strategic Trading in a women across the past century al­ ,. reduce the growth of wages and Two-Sided Foreign lows us to observe the choices ." living standards for natives while Exchange Auction each made, and determine how the increasing their unemployment? Linda Goldberg and constraints facing college women Rafael Tenorio have loosened over time. Improving the NBER Working Paper No. 5187 No cohort of college graduate Accessibility of the July 1995 women in the past had a high suc­ NBER's Historical Data International Finance and cess rate in combining family and Daniel R. Feenberg and Macroeconomics career. Cohort I (graduating c. Jeffrey A. Miron The microstructure chosen for 1910) had a 50 percent rate of NBER Working Paper No. 5186 foreign exchange markets can in­ childlessness. Cohort III (graduat­ July 1995 fluence trading volumes and equi­ ing c. 1955) had a high rate of JEL Nos. C82, E32, N12 librium exchange rates. With childbearing, but initially low labor Economic Fluctuations, International emerging markets and developing force participation. Cohort IV Finance and Macroeconomics, countries increasingly using two­ (graduating c. 1972) provides the Monetary Economics sided auctions, the choice of the most immediate guide for today's During the early years of its ex­ discrete "tatonnement" auction cre­ college women, and is close to the istence, the National Bureau of ates incentives for strategic behav­ end of its fertility history. It is also Economic Research assembled an ior among market participants. We a cohort that can.be studied using extensive dataset on all aspects of theoretically predict strategic un­ the National Longitudinal Survey of the pre-World War II macroecono­ derevelation of demand or supply Young Women. my. Until 1978, this dataset existed positions that are supported em­ In 1991, when the group was only on the handwritten sheets to pirically with detailed data from a between 37 and 47 years old, 28 which the early NBER researchers rare example of a tatonnement percent of the college graduate copied the data from original sour- market: the Moscow Interbank Cur­ (white) women had yet to have a ces. In 1978, the Inter-University rency Exchange. Our results also ftrst birth. An estimated 24 percent ) Consortium for Political and Social are consistent with findings from to 33 percent of the college gradu-

NBER ReporterFall 1995 45. ate women in the sample had a ca­ Why Are Retirement Rates the broader economy of defending • reer. Thus, only 13 percent to 17 So High at Age 65? an exchange rate peg can be very . percent of the group achieved Robin L Lumsdaine, high. We illustrate the dynamic in­ "family and career" by the time it James H. Stock, and terplay between credibility and was about 40 years old. Among David A. Wise commitment by describing the those who attained career, 50 per­ NBER Working Paper No. 5190 1992 Swedish and British crises cent were childless. Cohort N con­ July 1995 and the 1994-5 Mexican collapse. tains a small group of women who JEL Nos. J14, J26 We also discuss the small number have combined family with career, Aging, Labor Studies of successful "fixers." but for most the goal remains elusive. In most datasets of labor force USing Social Security Data Taxation and Corporate participation of the elderly, retire­ on Military Applicants to Investment: The Impact ment rates are particularly high at Estimate the Effect of of the 1991 Swedish age 65. While there are numerous Voluntary Military Service Tax Reform economic reasons why individuals on Earnings AlanJ. Auerbach, may choose to retire at 65, empiri­ Joshua D. Angrist Kevin Hassett, and cal models that have attempted to NBER Working Paper No. 5192 Jan Sodersten explain the spike at age 65 have July 1995 NBER Working Paper No. 5189 met with limited success. Interpret­ JEL No. J31 July 1995 ed another way, while many mod­ Labor Studies els would predict a jump in the JEL Nos. H32, E62 This study uses Social Security Public Economics hazard rate at age 65, the magni­ tude of the spike indicates exces­ data on the earnings of applicants In 1990, the government of Swe­ sive response, given the economic to the all-volunteer military forces to compare the earnings of veter- den introduced a major tax reform considerations that retirees typically ans of the armed forces with the to take effect in 1991. Prior to the face. This paper considers the puz­ earnings of military applicants who • legislation, the Swedish system was zle of why retirement rates are so did not enlist. I present matching, so complex that the likely effects high at age 65, and explores a vari­ regression, and instrumental vari- of the reform on incentives to in­ ety of explanations. vest were unknown. ables eN) estimates. The matching In this paper, we draw on So­ and regression estimates control for The Mirage of Fixed most of the characteristics used by dersten (989) and Auerbach and Exchange Rates Hassett (992) to derive an expres­ the military to select qualified ap­ Maurice Obstfe1d and plicants from the pool of military sion for the user cost of capital that Kenneth Rogoff captures the essential features of applicants. The N estimates exploit NBER Working Paper No. 5191 an error in the scoring of exams the Swedish tax code both before that were used by the military to and after the reform. We estimate July 1995 JEL Nos. F31, E42, E52 screen applicants between 1976 the model for investment in equip­ International Finance and and 1980. All of the estimates sug- ment, and find that the responsive­ gest that soldiers who served in the ness of Swedish firms to the user Macroeconomics, Monetary Economics early 1980s were paid considerably cost is quite similar to what is This paper discusses the pro­ more while they were in the mili- found for the United State». Finally, found difficulties of maintaining tary than comparable civilians. Mili- we use our model and our esti­ fixed exchange rates in a world of tary service also appears to have mates to assess the effects of the expanding global capital markets. led to a modest (less than 10 per- 1991 reform. We find that the im­ Contrary to popular wisdom, mon­ cent) increase in the civilian earn- pact of the reform on investment etary authorities in industrialized ings of nonwhite veterans, while was probably minor, and had little countries easily have the resources actually reducing the civilian earn- to do with the contemporaneous to defend exchange parities against ings of white veterans. Most of the sharp drop in investment. virtually any private speculative at­ positive effects of military service tack. But if their commitment to on civilian earnings appear to be use those resources lacks credibili­ attributable to improved employ- ty with markets, then the costs to ment prospects for veterans.

46. NBER Reporter Fall 1995 ) The-Nature of theoretical relationship between Choosing a Dictator: Precautionary Wealth wealth and uncertainty in a buffer­ Bureaucracy and Welfare Christopher D. Carroll and stock model of saving. Then we es­ in Less-Developed Polities Andrew A. Samwick timate that equation using PSID James E. Rauch NBER Working Paper No. 5193 data; we find strong evidence that NBER Working Paper No. 5196 July 1995 households engage in precaution­ July 1995 JEL Nos. D91, E21 ary saving. Finally, we simulate the JELNo. D73 Aging, Economic Fluctuations, wealth distribution that would pre­ International Trade and Investment Public Economics vail if all households had the same uncertainty as the group with the Recent work in the sociology of This paper uses the Panel Study least uncertainty. We fmd that be­ economic development has empha­ of Income Dynamics to provide tween 39 and 46 percent of wealth sized that establishing a profession­ some of the first direct evidence in our sample is attributable to differ­ al government bureaucracy in place that wealth is systematically higher entials in uncertainty across groups. of political appointees is important for consumers whose income is to an institutional environment in more uncertain. However, the ap­ Alcohol Poli~ies and which private enterprise can flour­ parent pattern of precautionary Highway Vehicle Fatalities ish. This paper focuses on the role saving is not consistent with a stan­ Christopher J. Ruhm of internal promotion in bringing dard parameterization of the life­ NBER Working Paper No. 5195 to power individuals who vahle im­ cycle model, in which consumers July 1995 posing their preferences toward are patient enough to begin saving JEL No. 118 collective goods on the public for retirement early in life: wealth Health Economics more than income. Such individu­ is estimated to be less sensitive to als restrain the corruption of their uncertainty in permanent income This study investigates the im­ subordinates by using tax revenue than that model implies. Instead, pact of beer taxes and a variety of to implement their preferences. our results suggest that over most alcohol-control policies on motor Within this hierarchical framework, ).. of their working lifetime, con­ vehicle fatality rates. I use fIXed-ef­ I investigate the effects of varying sumers behave in accordance with fect models with data for the 48 the compensation levels of subordi­ the "buffer-stock" models of saving contiguous states from 1982 nates and of recruiting them ac­ described in Carroll (992) or Dea­ through 1988. It is also necessary cording to merit to control adequately for grassroots ton (1991), in which consumers efforts to reduce drunk driving, the hold wealth principally to insulate enactment of other laws that oper­ Real Effects of Exchange consumption against near-term ate simultaneously to reduce high­ Rate-Based Stabilization: fluctuations in income. way fatalities, and the economic An Analysis of Competing Theories How Important Is conditions existing at the time of the legislation. In the specifications Sergio T. Rebelo and Carlos A. Vegh Precautionary Saving? that I prefer, most of the regula­ Christopher D. Carroll and tions have little or no impact on NBER Working Paper No. 5197 Andrew A. Samwick traffic mortality. By contrast, higher July 1995 NBER Working Paper No. 5194 beer taxes are associated with re­ JEL No. F41 July 1995 ductions in crash deaths; this result Growth, International Finance and JEL Nos. D91, E21 is relatively robust across specifica­ Macroeconomics Aging, Economic Fluctuations, tions. These findings suggest that We use a unified analytical Public Economics further regulatory action has a lim­ framework to assess, both qualita­ We estimate what fraction of the ited ability to reduce drunk driving, tively and quantitatively, the rele­ wealth of a sample of Panel Study but higher alcohol taxes can playa vance of the different hypotheses of Income Dynamics (PSID) re­ potentially significant role. that have been proposed to ex­ spondents is held because some plain the real effects of exchange households face greater uncertainty rate-based stabilizations. The four of income than others. First we de- major hypotheses we analyze are: ~ rive an equation characterizing the 1) the supply-side effects assodat-

NBER ReporterFall 1995 47. ed with a decline in inflation; 2) July 1995 An Empirical Analysis of the perception that the exchange JEL No. 031 Alcohol Addiction: Results rate peg is temporary; 3) the fiscal Productivity from the Monitoring-the- adjustments that tend to accompany Future Panels • Scientists who make break- the peg; and 4) the existence of through discoveries can receive Michael Grossman, nominal rigidities in wages or prices. Frank]. Chaloupka, and above-normal returns to their intel- Ismail Sirtalan lectual capital, depending upon the NBER Wolking Paper No. 5200 On the Ills of Adjustment degree of natural excludability- July 1995 Ricardo]. Caballero and that is, whether necessary tech- JEL No. I10 Mohamad L Hammour niques can be learned through Health Economics NBER Wolking Paper No. 5198 written reports, or instead require July 1995 hands-on experience with the dis- This paper aims to refine and JEL Nos. E10, FOO, J60 covering scientists or those trained enrich the empirical literature on Economic Fluctuations by them in their laboratory. Priva- the sensitivity of alcohol consump- We analyze market impediments tizing discoveries, then, only re- tion and excessive consumption to to the process of structural adjust- quires selecting trusted others as differences in the prices of alco- ment, focusing on incomplete-con- collaborators, most often scientists holic beverages. Mainly, we incor- tract inefficiencies in transactions working in the same organization. porate insights provided by a mod- between workers and firms that Within organizational boundaries, el of rational addictive behavior render the quasi-rents from "specif- incentives become aligned based that emphasizes the interdepen- ic" investment appropriable. During on repeat and future exchange, dency of past, current, and future adjustment, there is a depressed coupled with third-party monitor- consumption of an addictive good. rate of creation of the new produc- ing and enforcement. This study uses a U.S. panel with tive structure and excessive de- We find that high-value intellec- members ranging in age from 17 struction of the old structure, lead- tual capital paradoxically predicts through 27. Since the prevalence of ing to an employment crisis. More- both a larger number of collabora- alcohol dependence and abuse is over, appropriability weakens the tors and more of that network con- highest in this age range, addictive incentives for extensive restructur- tained within the same organiza- models of alcohol consumption ing, and results in a "sclerotic" pro- may be more relevant to this sam- • tion. Specifically, same-organiza- ductive structure. An adequate pIe than to a representative sample managed-adjustment program com- tion collaboration pairs are more likely when the value of the intel- of the population of all ages. bines vigorous creation incentives We find that alcohol consump- in the expanding sector with mea- lectual capital is high: both are tion by young adults is addictive, in sures to support employment in the highly productive "star" scientists, the sense that increases in past or contracting one. In contrast, the both are located in top-quality bio- future consumption cause current common prescription of gradualism science university departments, or consumption to rise. The positive does not act as an effective "syn- both are located in a firm (higher and significant future effect on con- chronizer" of creation and destruc- ability to capture returns). In con- sumption is consistent with the hy- tion, because it can reduce destruc- trast, collaboration across organiza- tion only by also reducing an alrea- tion boundaries is related negative- pothesis of rational addiction, and dy depressed creation rate. ly to the value of intellectual capi- inconsistent with the hypothesis of tal and positively to the number of myopic addiction. The long-run Collaboration Structure times the star scientist has moved. elasticity of consumption with re- and Information Dilemmas Organizational boundaries act as spect to the price of beer is ap- prOximately 60 percent larger than in Biotechnology: "information envelopes." The more Organizational Boundaries valuable the information produced, the short-run price elasticity, and as Trust Production the more limited is its dissemina- twice as large as the elasticity that ignores addiction. Lynne G. Zucker, tion. In geographic areas where a Michael R. Darby, higher proportion of coauthor pairs Marilynn B. Brewer, and come from the same organization, Yusheng Peng diffusion to new collaborators is NBER Wolking Paper No. 5199 retarded. •

48. NBER Reporter Fall 1995 ) The Incidence of a The Decline of workers expanded in the manufactur- Firm-Varying Payroll Tax: Noncompeting Groups: ing sector. We compare this his tori- The Case of Changes in the Premium cal period of narrowing wage differ- Unemployment Insurance to Education, 1890 to 1940 entials in the face of technological Patricia M. Anderson and Claudia Goldin and progress in the office to our current Bruce D. Meyer Lawrence F. Katz period of widening differentials. NBER Wolking Paper No. 5201 NBER Working Paper No. 5202 August 1995 August 1995 What Can Explain the JEL Nos. H22, J65 JEL Nos. JO, NO Apparent Lack of Labor Studies, Public Economics Development of the American International Consumption Economy, Labor Studies In this paper, we examine the Risksharing? Karen K Lewis common case of a firm-varying tax Between 1890 and the late 1920s used to finance a fringe benefit. the premium to high school educa- NBER Working Paper No. 5203 While our data come from the ex- tion declined substantially for both August 1995 perience-rated unemployment in- men and women:. In 1890 ordinary International Finance and surance system, the differential office workers, • whose positions Macroeconomics, Asset Pricing treatment of firms (including spe- generally required a high school di- Recent research in international cial considerations for small busi- ploma, earned almost twice what business cycles based on complete ness) under mandated benefits production workers did. But by the markets has found that correlations laws leads to firm-varying costs late 1920s, office workers earned in international consumption are analogous to those occurring with about one and a half times as lower than the standard risk sharing experience-rated taxes. Our theo- much as production workers. The implications of these models would retical model highlights the impor- premium earned by female office predict. In this paper, I use regres- tance of this variation in taxes or workers, male office workers, and sion tests to determine which of costs, both within and across markets. "male office workers plus supervi- two different types of explanations ) We examine annual changes in sors" fell by about 30 percent. are valid. First, I consider whether both average earnings and employ- Several factors operated in tan- nonseparabilities between trad- ment within firms and earnings of dem to narrOw differentials to edu- abIes and nontradable leisure or individual workers at the same cation: the supply of high school goods can explain the puzzle. Sur- firm. This method removes unmea- graduates relative to those without prisingly, nonseparabilities explain sured firm and worker characteris- high school degrees increased by only a tiny fraction of the variation tics, thus avoiding the omitted vari- 16 percent from 1890 to 1910, but in the consumption of tradables able bias that has plagued past by 40 percent from 1910 to 1920, across countries. Furthermore, I re- work on incidence and compensat- and by 50 percent from 1920 to ject the hypothesis of risksharing in ing differentials. Our results sug- 1930. Immigration restriction was tradables. gest that most of the market-level another factor, but was dwarfed by Second, I examine the effects of tax is borne by the worker. How- the expansion of high schools; re- capital market restrictions on ag- ever, this does not imply that there duced immigrant flows explain just gregate consumption risksharing are no effects of the tax on em- one-eighth of the relative increase by countries. Risksharing definitely ployment. Rather, we find that in- in the supply of educated workers. does not occur for countries facing dividual firms can pass along only The impact of rapidly increasing more severe capital market restric- a small share of the within-market supplies of high school educated tionS, nor does it occur among the differences in the tax they face, workers was reinforced by techno- unrestricted group of countries. leading to substantial employment logical changes in the office that Therefore, risksharing does not ap- reallocation across firms. enabled the substitution of educat- pear to be resolved by either type ed workers and machines for the of explanation alone. However, exceptionally able. when I allow for both nonsepara- The premium to high school bilities and certain market restric- graduation, rather than declining tions, then risksharing among urue- further in the 1930s, leveled off as stricted countries appears to exist. ~ the demand for high school educated This suggests that a combination of

NBER RepotterFalll995 49. these two effects may be necessary and more countries pass through sessments (a form of taxation) to to explain consumption risksharing the development phase, or will it state guarantee funds (called "sol- across countries. be controlled? Intuitively, if "clean vency funds") in order to protect air" is a normal good, we would the policyholders of the failed com- • Price Level Determinacy expect that societies might be "self- pany. We estimate the costs to the Without Control of a regulating" in the sense that as in- guarantee funds of resolving insol- Monetary Aggregate come increases, pollution controls vencies of P and C insurance com- MUchaelVVoodford also increase. However, this intu- panies. We find that the total net NBER Wolking Paper No. 5204 ition is somewhat misleading, be- costs (that is, payments by the fund August 1995 cause the presence of external ef- less recoveries by the fund) of re- Economic Fluctuations, fects is an essential feature of envi- solving insolvencies are remarkably Monetary Economics ronmental regulation. high. We estimate that the mean ra- This paper describes a growth tio of net-costs-to-assets is approxi- I show that the price level re- model in which pollutants are in- mately one; this implies that insol- mains determinate even in the case ternal to a jurisdiction. We develop vent companies have roughly twice of two kinds of radical money sup- a model of the joint determination as many liabilities as assets when ply endogeneity-an interest rate of the rate of development of the they fail. Our cost estimate for re- peg by the central bank, and a economy through market interac- solving insurance company insol- "free banking" regime-commonly tions and the extent of pollution vencies is roughly three times the supposed to imply loss of control regulation through collective deci- comparable estimate for banks. We of the price level. Determination of sionmaking. We show that depend- also find that the ratio of net-costs- the price level under such regimes ing on the collective decisionmak- to-assets tends to be higher for can be understood in terms of a ing mechanism in place, the time small firms, poorly capitalized "fiscal theory of the price level," path of pollution can display an in- firms, firms writing Significant pre- according to which the equilibrium verted U shape, a "sideways mir- miums in long tail lines, and firms price level makes the real value of rored" S, or an increasing (but that fail because of disasters. The nominally denominated govern- bounded) level over time. resolution of insolvencies is typical- ment liabilities equal to the present This paper contributes to the lit- ly quick: more than 60 percent of all value of expected future govern- erature on both the large differ- costs to the fund for a given insolven- • ment budget surpluses. I also out- ences in income per capita across cy occur within two years, and more line the application of the fiscal countries and the discrepancies in than three-quarters of total costs theory of the price level to exoge- their growth rates. It shows that by occur within three years. However, nous money regimes. relying on collective decisionmaking firms with a high proportion of pre- mechanisms to choose policies, the miums in long tail lines take much A Positive Model of Growth dynamics of convex models can re- longer to resolve insolvencies. and pollution Controls semble those usually ascribed to Larry E. Jones and models of multiple equilibriums. Engines of Growth: Rodolfo E. Manuelli Domestic and Foreign NBER Working Paper No. 5205 Property and Casualty Sources of Innovation August 1995 Solvency Funds as a Jonathan Eaton and JEL Nos. E6, HI, 01 Tax 3;nd Social Samuel Kortum Growth Insurance System NBER Working Paper No. 5207 James Bohn and BrianJ. Hall August 1995 The most recent addition to the NBER Working Paper No. 5206 Growth, International Trade and "economics of gloom" concerns the August 1995 Investment, International Finance and interplay between income and en- JEL No. G22 Macroeconomics, Productivity vironmental degradation. The main Public Economics question is whether continued en- We examine productivity growth vironmental degradation is a neces- When a Property and Casualty since World War II in the five lead- sary part of the process of industri- (P and C) insurance company be- ing research economies: West Ger- alization. Will pollution continue to comes insolvent, solvent insurance many, France, the United King- increase without bound as more companies are forced to pay as- dom, Japan, and the United States. • 50. NBER Reporter Fall 1995 ) Data on the capital-output ratio workers on the job even at 3 a.m. growth falls sharply during discrete suggest that these countries grew 2) Such work is performed mostly high inflation crises, then recovers as they did because they were able by people who are not shift work- surprisingly strongly after inflation to adopt more productive techno 1- ers. 3) Work at these times is inferi- falls. The decline in growth during ogies, not because of capital deep- or, in that it is performed dispro- crisis, and recovery of growth after ening per se. We present a multi- portionately by people with little crisis, tend to average out to close country model of technological in- human capital. 4) Minority workers to zero (even slightly above zero); novation and diffusion that implies in the United States and the for- hence there is no robust cross-sec- that, for a wide range of parameter eign-bom in Germany are especial- tion correlation. Our fmdings could values, countries converge to a ly likely to work at these undesir- be consistent either with trend sta- common growth rate, with relative able times. 5) Evening and night tionarity of output, in which infla- productivities depending on the work is least likely in large metro- tion crises are purely cyclical phe- speed with which countries adopt politan areas. 6) Spouses tend to nomena, or with models in which technologies developed at home work at the same time of the day. crises have a favorable long-run and abroad. Using parameter val- 7) However, yatIng children break purgative effect. Our findings do ues that fit a cross section of data down the joint pming of spouses' not support the view that reduction on productivity, research, and pat- work, with the burden of evening of high inflation carries heavy enting, we simulate the growth of and night work falling dispropor- short-to-medium-run output costs. the five countries, given initial pro- tionately on working mothers. ductivity levels in 1950 and re- These findings demonstrate the The Political Economy of search efforts in the four subse- gains to basing the analysis of Branching Restrictions quent decades. Based on plausible work and leisure on data des crib- and Deposit Insurance: assumptions about "technology ing instantaneous time use. A Model of Monopolistic j gaps" that existed among these Competition Among Small I countries in 1950, we can explain Inflation Crises and and Large Banks I j their growth experiences quite suc- Long-Run Growth Nicholas Economides, cessfully. Specifically, the simula- Michael Bruno and R Glenn Hubbard, and 1 -- tions capture the magnitude of the william. Easterly Darius palla I slowdown in German, French, and 1 NBER Working Paper No. 5209 NBER Working Paper No. 5210 I I Japanese productivity growth, and August 1995 August 1995 I the relative constancy of u.K. and JEL Nos. 011, 040, E63, E31 JEL Nos. G2, L5 U.S. growth. International Finance and Corporate Finance, Macroeconomics Monetary Economics Who Works When? Evidence from the United Recent literature suggests that This paper suggests that the in- States and Germany long-run averages of growth and troduction of bank branching re- strictions and federal deposit insur- Daniel S. Hamermesh inflation are only weakly correlat- ed, and such correlation is not ro- ance in the United States likely was NBER Worldng Paper No. 5208 bust to exclusion of extreme infla- motivated by political considera- August 1995 tion observations. Inclusion of tions. Specifically, we argue that JEL No. J22 time-series panel data has im- these restrictions were instituted for Labor Studies proved matters, but an aggregate the benefit of the small, unit banks This study uses data for the parametric approach remains in- that were unable to compete effec- United States from the May 1991 conclusive. We propose a nonpara- tively with large, .multi-unit banks. Current Population Survey and for metric definition of high inflation We analyze this "political hypo the- Germany from the 1990 wave of crises as periods when inflation is sis" in two steps. First, we use a the Socioeconomic Panel to ana- above 40 percent annually. Exclud- model of monopolistic competition lyze when people work during the ing countries with high inflation between small and large banks to day and week. The evidence crises, we find no evidence of any examine gains to the former group shows: 1) Work in the evenings or consistent relationship between from the introduction of branching at night is quite common in both growth and inflation at any fre- restrictions and government-spon- ; countries, with around 7 percent of quency. However, we find that sored deposit insurance. We then

NBER RepotterFall 1995 51. fmd strong evidence for the politi­ rate. Furthermore, increased prod­ Preference Parameters • cal hypothesis by examining the uct variety, brought about by in­ and Behavioral voting record of Congress. creases in market size, might re­ Heterogeneity: An duce the returns to improving Experimental Approach Growth Without product quality, paradoxically low­ in the Health and Scale Effects ering an economy's growth rate Retirement Survey Alwyn Young while increasing the total resources Robert B. Barsky, NBER Working Paper No. 5211 devoted to Rand D. Miles S. Kimball, August 1995 F. Thomas Juster, and Matthew D. Shapiro Growth NBER Working Paper No. 5213 An increase in the size (scale) of Does Monetary Policy August 1995 an economy increases the total Affect Real Economic JEL Nos. D81, D91, 042 quantity of rents that can be cap­ Activity? Why Do We Still Asset Pricing, Economic Fluctuations, tured by successful innovators. In Ask This Question? Monetary Economics equilibrium, this should lead to a Benjamin M. Friedman rise in innovative activity. Conven­ Individuals' preferences underly­ NBER Working Paper No. 5212 tional wisdom and the theoretical ing most economic behavior are AUgust 1995 predictions of models of endoge­ likely to display substantial hetero­ JEL No. E50 geneity. This paper reports on di­ nous innovation suggest that this Monetary Economics increased research effort should rect measures of preference param­ lead to more rapid growth . .As not­ The predominant weight of the eters relating to risk tolerance, time ed by Jones (1993), this prediction existing evidence suggests that the preference, and intertemporal sub­ is at odds with the postwar experi­ effects of monetary policy on real stitution. These experimental mea­ ence of the Organization for Eco­ economic' activity are systematic, sures are based on survey respon- nomic Cooperation and Develop­ significant, and sizable. Yet ques­ dents' choices in hypothetical situa- • ment, where the growth of the tions remain, both about individual tions. The questions are construct- market indeed has led to an in­ empirical results and, more broad­ ed with as little departure from the creased Rand D effort that, how­ ly, about the different methodologi­ theorist's concept of the underlying ever, has been translated into stag­ cal approaches that researchers parameter as possible. nant or declining growth rates. have used to investigate these ef­ The individual measures of pref­ Drawing upon the remarkable in­ fects. This paper addresses the con­ erence parameters display substan­ sights of the museum curator S. C. ceptual issues that account for our tial heterogeneity. The majority of Gilfillan (1935), this paper modifies continuing to ask whether mone­ respondents fall into the least risk­ models of endogenous innovation tary policy has real effects even tolerant group, but a substantial to allow for the possibility that a though, at a certain level, we do minority display higher risk toler­ rise in the profitability of innova­ "know" the answer. The paper's ance. The individual measures of tive activity could lead to an in­ overview of theory and evidence intertemporal substitution and time creased variety of differentiated so­ suggests that much of the explana­ preference also display substantial lutions to similar problems. An in­ tion for the continuing tug-of war heterogeneity. The mean risk toler­ creased variety of technologies (for between research findings and sub­ ance is 0.25; the mean elasticity of example, an increase in the num­ sequent questions in this area lies intertemporal substitution is 0.2. ber and types of contraceptives) in two sets of limitations, one re­ Estimated risk tolerance and the will increase the level of utility of flecting how economics conceptu­ elasticity of intertemporal substitu­ the average consumer. However, if alizes behavioral processes and one tion are essentially uncorrelated continued improvement of this in­ reflecting how economics draws in­ across individuals. creased variety of technologies re­ ferences from observed data. Because the risk tolerance mea­ quires increased research input, sure is obtained as part of the main then a rise in the scale of the mar­ questionnaire of a large survey, it ket could raise the equilibrium can be related to -a number of eco­ quantity of Rand D, without in­ nomic behaviors. Measured risk creasing the economy's growth tolerance is related positively to a

52. NBER Reporter Fall 1995 ) number of risky behaviors, includ­ International Finance and mortality. The incentive to invest in ing smoking, drinking, failing to Macroeconomics, International Trade prevention against one cause of have insurance, and holding stocks and Investment death depends positively on the level of survival from other causes. rather than Treasury bills. Although ,tending McCallum's (1995) re­ This means that a specific public measured risk tolerance explains sult, based on a gravity model of health intervention has benefits only a small fraction of the varia­ 1988 trade flows, that a typical Ca­ other than the direct medical re­ tion of the studied behaviors, these nadian province trades 22 times duction in mortality: it affects the estimates provide evidence about more with other provinces than incentives to fight other diseases, the validity and usefulness of the with U.S. states of similar size and so the overall reduction in mortality measures of preference parameters. distance, this paper asks how Que­ in general will be larger than that bec trade patterns compare with predicted by the ditect medical ef­ Fiscal Expansions and those of other provinces. The re­ fects. We discuss evidence of these Fiscal Adjustments in sults, based on revised data for cross-disease effects by using data OECD Countries 1988, 1989, and 1990, show that on neonatal tetanus vaccination Alberto Alesina and while the typical province trades through the Expanded Programme Roberto Perotti more than 20 times as much with on Immunization of the World NBER Working Paper No. 5214 other provinces as with compara­ Health Organization. August 1995 ble u.s. states, the multiple is even Monetary Economics greater for\, Quebec. Thus trade be­ tween Quebec and the United An Adverse-Selection This paper considers budget ex­ States appears to be an even less Model of Bank Asset and pansions and adjustments in Orga­ viable alternative to interprovincial Liability Management nization for Economic Cooperation trade for Quebec than it is for the with Implications for and Development countries in the rest of Canada. The implications of the Transmission of last three decades. Our main results these results for international eco­ Monetary Policy • are: 1) on average, fiscal expan­ nomics are considerable, as they Jeremy c. Stein , sions are the result of increases in show that trade linkages within a NBER Working Paper No. 5217 expenditures, particularly of trans­ national economy are far greater August 1995 fer programs, while contractions than previously imagined. If those Corporate Finance, Monetary typically are caused by tax increas­ results are confirmed, they imply Economies, Economic Fluctuations es; 2) however successful (that is, that the fabric of national econo­ This paper develops a model of long lasting), a minority of the total mies is far tighter than that of the bank-asset-and-liability manage­ rely primarily on reduction of gov­ global trading system, even for ment, based on the idea that infor­ ernment wages and employment countries operating without sub­ mation problems make it difficult and cuts in transfer programs; 3) stantial trade barriers. for banks to raise funds other than even major successful fiscal adjust­ with insured deposits. The model ments do not seem to have reces­ Disease Complementarities can be used to address the ques­ sionary consequences, on average; and the Evaluation of tion of how monetary policy and 4) different types of govern­ Public Health Interventions works. One effect it captures is that ments show different degrees of william H. Dow, Jessica Holmes, when the Fed reduces reserves, success at implementing successful Tomas Philipson, and this tightens banks' finanCing con­ fiscal adjustment, with coalition Xavier Sala-i-Martin straints, thereby leading to a cut­ governments showing the worst NBER Working Paper No. 5216 back in bank lending; this is the performance. August 1995 "bank ,lending channel" in action. JEL Nos. 10, 11 However, in addition to providing Do National Borders Growth, Health Economics a specific set of microfoundations Matter for Quebee's Trade? This paper provides a theoreti­ for the lending channel, the model John F. Helliwell cal and empirical investigation of also yields a novel account of how NBER Working Paper No. 5215 the positive complementarities be­ monetary policy affects interest rates in the bond market. ,~ August 1995 tween disease-specific policies in­ .., JEL Nos. F14, F15 troduced by competing risks of

NBER RepotterFall 1995 53. High-Income Families and August 1995 Openness and Growth: the Tax Changes of the JEL Nos. F31, F33 A Time-Series, Cross- 1980s: The Anatomy of International Finance and Country Analysis for Macroeconomics • Behavioral Response Developing Countries Joel B. Slemrod This paper addresses the issue Ann Harrison NBER Wolking Paper No. 5218 of whether regimes of fixed ex- NBER Working Paper No. 5221 August 1995 change rates are a mechanism for August 1995 JEL No. H2 shifting volatility intertemporally. JEL Nos. F43, 047 Public Economics Using a panel of data covering 20 International Trade and Investment industrialized countries from 1959 The relative income gains of the through 1993, I examine the vola- This paper draws together a va- affluent after the passage of the tilities of a host of real and mone- riety of measures of openness to Tax Reform Act of 1986 (TRA86), tary variables. Graphical and statis- test its association with growth. Al- which sharply lowered tax rates at tical examination of the periods though the correlation across dif- high income levels, are overstated around 33 flotations and 81 devalu- ferent types of openness is not al- by comparing cross-sectional slices ations reveals little evidence of sig- ways strong, there is generally a pos- using concurrent income defini- nificant increases in volatility fol- itive association between growth tions. Nevertheless, they are large. lowing these events. and different measures of open- Although an index of the demand- ness. The strength of the associa- side factors affecting inequality The Seamless World: tion depends on whether the spec- throughout the income distribution A Spatial Model of ification uses cross-sectional or can explain much of the increased International Specialization panel data (which combines cross high-income concentration until Paul R. Krugman and section and time series). For indus- 1985, it cannot adequately explain Anthony J. Venables trializing countries, which have ex- the post-TRA86 spurt. Thus, TRA86 NBER Working Paper No. 5220 hibited significant fluctuations in is likely to have been a principal August 1995 trade regimes over time, long-run cause of the large increase in the JEL Nos. F1, F12, F15 averages may not serve as very reported personal income of the af- International Trade and Investment, meaningful indicators of policy. fluent. A close look at the sources International Finance and • of the post-1986 increases in the re- Macroeconomics The Language Ability of ported individual income of high- This paper is an effort to do in- US. Immigrants: income households suggests that Assimilation and much is caused by shifting of in- ternational trade theory without mentioning countries. Nearly all Cohort Effects come-for example, from the cor- Geoffrey Carliner porate tax base to the individual models of the international econo- NBER Working Paper No. 5222 tax base-and not income creation, my assume that trade takes place August 1995 such as additional labor supply. between nations or regions that are This distinction is critical, because themselves dimensionless points. JEL Nos.J15,J24, F22 Labor Studies knowing how much the reported We develop a model in which eco- nomic space instead is assumed to individual income of a particular This paper uses data from the be continuous, and in which this group of people changes in re- "seamless world" spontaneously or- 1980 and 1990 u.s. Census of Pop- sponse to a tax change is not suffi- ganizes itself into industrial and ag- ulation to examine the English lan- cient for evaluating adequately its ricultural zones because of the ten- guage skills of natives and immi- revenue consequences, incidence, sion between forces of agglomera- grants. The first main finding is that and efficiency. tion and disagglomeration. One lack of fluency in spoken English is might expect such a model to be rare among native-born Americans. After the Deluge: Do Fixed analytically intractable, but we are In 1990, 98.4 percent of natives Exchange Rates Allow able to gain considerable insight aged 18 to 64 reported speaking Intertemporal Volatility through a combination of simula- only English or speaking it very Trade-Offs? tions and an analytical approach well. Among native-born children Andrew K. Rose originally suggested in a biological of ethnic groups who have come NBER WOlking Paper No. 5219 context by Alan Turing. to the United States in large num- •

54. NBER Repon-erFall 1995 -~ bers during the past 30 years, such Factors Affecting Labor The Effect of New Jersey's as Hispanics and East Asians, a Supply Decisions and Minimum Wage Increase substantial fraction were not fluent Retirement Income on Fast-Food Employment: when they entered grade school, Robin L Lumsdaine A Reevaluation Using but at most 3 to 5 percent of teen­ NBER Working Paper No. 5223 Payroll Records agers and adults in these groups August 1995 David Neumark and reported speaking English poorly JEL Nos. J14, J26 William WaScher or not at all. Aging NBER Working Paper No. 5224 Second, the vast majority of im­ August 1995 Recent policy has focused on al­ migrants speak English well. In JEL Nos. J23, J38 leviating poverty among the elder­ 1990, only a quarter of immigrants Labor Studies reported speaking English poorly ly, and has met with varying de­ or not at all, although more than grees of success. Some subsets of We reevaluate the evidence from half of Mexicans and one-third of the elderly population have gained Card and Krueger's (CK's) 1994 immigrants from other non-English­ at the expense of others. One com­ New Jersey-Pennsylvania minimum speaking western hemisphere ponent of the pblicy debate has wage experiment, using new data coun-tries could not speak English been identifying which factors based on actual payroll records well. Although English skills might influence decisions about la­ from 230 Burger King, KFC, Wen­ improve with length of residence, bor force participation, and what dy's, and Roy Rogers restaurants in after 30 or more years in the effects these decisions will have on New Jersey and Pennsylvania. We United States over a quarter of retirement income and its adequa­ compare results using these payroll Mexican immigrants spoke English cy for a growing elderly popula­ data to those using CK's data, poorly or not at all. tion. While models of retirement which were collected by telephone Third, since the 1950s there has behavior are becoming increasingly surveys. We have two findings to been a trend decrease in the proba- sophisticated, most fail to capture report: '; bility of fluency (speaking only En­ key elements, such as expectations First, the data collected by CK ", glish or speaking it very well) and uncertainty. In part this is appear to indicate greater variation among new immigrants of about caused by the reduced-form nature in employment over the eight­ 0.1 percentage points per year, of policy experiments: parameters month period between their sur­ caused by the shift from European are estimated under a current poli­ veys than the payroll data do. For immigrants with strong English cy, and used to predict effects of. example, in the full sample, the skills to Latin American and East an alternative scenario. This ap­ standard deviation of employment Asian immigrants who arrive speak­ proach implicitly assumes that the change is three times as large in ing less English. On average, each only difference in the alternative CK's data as in the payroll data. additional year of residence in the setting is the change in policy, and Second, estimates of the em­ United States increases the proba­ does not account adequately for ployment effect of the New Jersey bility of fluency by 1.1 percentage endogeneity of decisions and re­ increase in the minimum wage points. An additional year of sponses to these changes. This pa­ based on the payroll data lead to schooling increases the probability per reviews factors affecting the la­ the opposite conclusion from what of fluency by about 5 percentage bor supply decision, their interac­ CK reached For comparable sets of points. Overall, women are slightly tions with and implications for sub­ more likely to be fluent than men, restaurants, differences-in-differ­ sequent retirement income, and ences estimates using CK's data im­ especially East Asian and European identifies important methods and women. Even after controlling for dif­ ply that the New Jersey minimum data requirements necessary to ferences in education, years since wage increase (of 18.8 percent) re­ model complicated dynamic be­ arrival, and other factors, large dif­ sulted in an employment increase havior more accurately. ferences in English skills by region of 17.6 percent relative to the Penn­ of origin remain. These differences sylvania control group, an elasticity seem to be associated more with of 0.93. In contrast, estimates based geographic distance from the Unit­ on the payroll data suggest that the ed States than with the source New Jersey minimum wage in­ country's per capita income or lin- crease led to a 4.6 percent decrease ; guistic distance from English. in employment in New Jersey rela-

NBER ReponerFalll995 55. tive to the Pennsylvania control dence suggests that trade reform has tion relates to the q-theory of in­ group. This decrease is statistically been accompanied by significant vestment, and how these options significant at the 5 percent level increases in investment inflows. affect the incentive to invest. Gen­ and implies an elasticity of employ­ erally, the option to expand re­ • ment with respect to the minimum Measuring the Intensity duces the incentive to invest, while wage of -0.24. of Competition in the option to disinvest raises it We Export Markets show how these options interact to The Effects of Trade Pinelopi K. Goldberg and determine the effect of uncertainty Policy Reform: What Do Michael M. Knetter on investment, how these option We Really Know? NBER Working Paper No. 5226 values change in response to shifts Ann Harrison and August 1995 of the distribution of future prof­ Ana Revenga International Trade and Investment itability, and how the q-theory and NBER Working Paper No. 5225 option pricing approaches to in­ August 1995 This paper develops an ap­ vestment are related. International Trade and Investment proach to measuring the intensity of competition in international mar­ Nonemployment and The magnitude of existing re­ kets. The method measures the de­ Health Insurance Coverage search on the effects of trade re­ gree of "outside" competition faced Jonathan Gruber and form is impressive. Yet economists by exporters located in one source Brigitte C. Madrian have not reached a clear consensus country from firms located outside NBER Working Paper No. 5228 on a number of important ques­ the source country. We use the elas­ August 1995 tions, including the labor market ticity of price and quantity to ex­ JEL Nos. H51, J64 impact of trade reform, the link­ change rate shocks, which shift the Health Care. Labor Studies, ages between trade and foreign di­ relative costs of producers from a Public Economics rect investment, and the relation­ particular source country, to calcu­ ship between trade and growth. In late our measure of outside compe­ Low rates of health insurance cov­ this paper we attempt to clarify tition. We estimate the measures erage among those who are without what we know about the relation­ using panel data on exports of u.s. a job have motivated consideration ship among trade reform, factor linerboard and German beer to a of policies to subsidize their pur­ • markets, and growth. variety of destination markets. The chase of insurance. But there is lit­ Although many studies have destination-specific panel data al­ tle evidence on the extent to which shown a positive relationship be­ low comparisons of outside com­ differentials in coverage between the employed and the nonem­ tween various measures of open­ petition across destination markets. ness and growth, many nagging ployed reflect the effects of job loss problems remain. Trade policy is versus merely different underlying Options, the Value of tastes for insurance. If it is the lat­ almost never measured using the Capital, and Investment most obvious indicators, for exam­ ter, then subsidies may not be suc­ Andrew B. Abel, ple tariffs. Further, many studies cessful in increasing the rate of in­ Avinash K. Dixit, surance coverage among the non­ are plagued by serious economet­ Janice C. Eberly, and employed Furthermore, subsidies that ric problems. The evidence on la­ Robert S. pindyck bor markets and trade reform is lower the costs of nonemployment NBER ~orking Paper No. 5227 may increase both the incidence less extensive. Based on fue stud­ August 1995 and the duration of joblessness. ies to date, it appears that the un­ Asset Pricing, Economic Fluctuations, employment and wage effects of Industrial Organization, Productivity We provide new evidence on trade reforms generally have been these issues by analyzing longitudi­ small. In this paper, we discuss the Capital investment decisions nal data on 25-54-year-old men for possibility that small wage and em­ must recognize the limitations on 1983-9. We have four interesting ployment responses are caused by the firm's ability to sell off or ex­ fmdings: First, even after modeling labor market regulations. We con­ pand capacity later. This paper differences in underlying tastes for clude with an analysis of the link­ shows how opportunities for future insurance, we find that the likeli­ ages between trade policies and expansion or contraction can be hood of insurance coverage drops foreign investment flows. Our evi- valued as options, how this valua- by roughly 20 percentage points

56. NBER Reporter Fall 1995 ·~ following separation from a job. drops in net worth. This suggests dren). However, there is evidence , Second, limited subsidization of the that insurers rely primaily on future of substantial bias caused by mea­ cost of insurance through state retained earnings to rebuild their surement error in the estimates of laws mandating continued access capital position. When property­ the effects of parents' socioeco­ to employer-provided health insur­ casualty insurers do go to the capi­ nomic status. In addition, Herrn­ ance for the nonemployed increas­ tal markets, we find that they do stein and Murray's measure of pa­ es by 6.7 percent the likelihood of not receive an unusually poor re­ rental SES fails to capture the ef­ having insurance but no job. Third, ception. In fact, the market price fects of important elements of fam­ these mandates also increase the reaction to equity issues appears to ily background (such as single-par­ number of individuals with spells be considerably less negative than ent family structure for youth at of nonemployment and the total for industrial issuers, but similar to age 14). As a result, their analysis amount of time spent jobless. Final­ that for banks and utilities. gives an exaggerated impression of ly, at least some of this increased the importance of IQ relative to nonemployment appears to be A Reanalysis of parents' SES, and relative to family spent in productive job search, as The Bell Curve background more generally. Esti­ the availability of continuation cov­ Sanders KoreruDan and mates based on a variety of meth­ erage is related to significant wage Christopher Wfuship ods, including analyses of Siblings, gains among those who separate NBER Working Paper No. 5230 suggest that parental family back­ from their jobs. August 1995 ground is at least as important as, JEL Nos. J3, J2, I3, J7 and may be much more important External Financing and Labor Studies than, IQ in determining social and economic success in adulthood. Insurance Cycles In The Bell Curve, Herrnstein Anne Gron and and Murray argue that a youth's in­ Deborah}. Lucas telligence CIQ) is a more important Beyond the Incidence of . NBER Working Paper No. 5229 determinant of social and economic Training: Evidence from a ~ August 1995 success in adulthood than is the so­ National Employers Survey , Corporate Finance cioeconomic status (SES) of his or Lisa M. Lynch and We explore the conjecture that her parents. They base this conclu­ Sandra E. Black the periodic episodes of high pri­ sion on comparisons of the effects NBER Working Paper No. 5231 ces and constrained supply in the of IQ score (measured at ages 15 August 1995 property-casualty insurance indus­ and 23) with the effects of an index JEL No. J24 try are the result of temporary of parents' SES taken from models Labor Studies shortages of capital. To do this we of economic status, marriage, wel­ This paper seeks to provide new look for increases in activities fare use, involvement in crime, as insights into how investments in aimed at increasing capital at these well as several outcomes for young school and post-school training are times: dividend cuts, repurchase children. However, reviewers of linked to employer workplace cuts, equity issues, and debt issues. The Bell Curve have questioned practices and outcomes. We use a By examining the market price re­ whether Herrnstein and Murray's unique nationally representative sponse to security issues, we also estimates of the effects of IQ are survey of U.S. establishments: the look for evidence that the costs of overstated by their use of a rather Educational Quality of the Work­ raising external capital are unusual­ crude measure of parents' SES. force National Employers Survey. ly high relative to other industries. Comparisons of siblings in the We go beyond simply measuring We find some evidence of chan­ Herrnstein and Murray sample, a the incidence of formal or informal ges in payout policy in the expect­ more comprehensive and accurate training to examine the determi­ ed direction, and also of an in­ way to control for family back­ nants of the types of training em­ creased volume of debt and equity ground, reveal little evidence that ployers invest in; the relationship issues after periods of low capaci­ these estimates of the effects of IQ between formal school and em­ ty. However, the total amount of score are biased because of omit­ ployer-provided training; who is capital obtained by security issues ted characteristics of family back­ receiving training; the links be­ or reduced payouts appears to be ground (with the possible excep­ tween investments in physical and ; small relative to the observed tion of outcomes. for young chil- human capital; and the impact that

NBER ReponerFall 1995 57. \

human capital investments have on important of them. In so doing, we necessary institution for the effi- • the productivity of establishments. find evidence of statistically and cient allocation of equity. We em­ We find that the smallest employ­ economically significant investment phasize this by providing an exam- ers are much less likely to provide responses to tax changes in 12 of ple of a banking system that per- formal training programs than em­ the 14 countries. forms as well. ployers from larger establishments. Regardless of size, those employers Stock Market Efficiency Risk Taking by Mutual who have adapted some of the prac­ and Economic Efficiency: Funds as a Response tices associated with what have Is There a Connection? to Incentives been called "high-performance Gary Gorton and James Dow Judith A. Chevalier and work systems" are more likely to NBER Working Paper No. 5233 Glenn D. Ellison have formal training programs. Es­ August 1995 NBER Working Paper No. 5234 pecially in the manufacturing sec­ JEL Nos. G14, G30 August 1995 tor, employers who have made Corporate Finance Corporate Finance, large investments in physical cap­ Industrial Organization ital, or who have hired workers In a capitalist economy prices This paper examines the agency with higher average education, are serve to equilibrate supply and de­ conflict between mutual fund in­ also more likely to invest in formal mand for goods and services, con­ vestors and mutual fund compa- training and to train a higher pro­ tinually changing to reallocate re­ . nies. Investors would like the fund portion of their workers. There are sources to their most efficient uses. However, secondary stock market company to use its judgment to significant and positive effects on maximize risk-adjusted fund re­ establishment productivity associat­ prices, often viewed as the most "informationally efficient" prices in turns. A fund company, however, ed with investments in human cap­ the economy, have no direct role in its desire to maximize its vahle ital. Those employers who hire in the allocation of equity capital, as a concern has an incentive to better-educated workers have ap­ since managers have discretion in take actions that increase the in­ preciably higher productivity. The determining the level of invest­ flow of investment. We use a semi­ impact of employer-provided train­ ment. What is the link between parametric model to estimate the ing differs according to the nature, stock price informational efficiency shape of the flow-performance re­ timing, and location of the employ­ and economic efficiency? lationship for a sample of growth er investments. We present a model of the stock and growth and income funds ob­ market in which: 1) managers have served from 1982-92. The shape of Tax Reforms and discretion in making investments the flow-performance relationship Investment: A Cross­ and must be given the right incen­ creates incentives for fund man­ Country Comparison tives; and 2) stock market traders agers to increase or decrease the Jason G. Cummins, may have important information riskiness of the fund, depending on A. Hassett, and Kevm that managers do not have about the fund's year-to-date return. Us­ R. Glenn Hubbard the value of prospective investment ing a new dataset of mutual fund NBER Working Paper No. 5232 opportunities. In equilibrium, infor­ portfolios that includes equity port­ August 1995 mation in stock prices will guide folio holdings for September and JEL Nos. H3, E2 investment decisions because man­ December of the same year, we Corporate Finance, Public EC9nomics agers ·will be compensated based show that mutual funds do alter We use firm-level panel data to on informative stock prices in the their portfolio riskiness between explore the extent to which fIxed future. September and December in a investment responds to tax reforms The stock market indirectly guides manner consistent with these risk in 14 OEeD countries. Previous investment by transferring two incentives. studies often have found that in­ kinds of information: information vestment does not respond to about investment opportunities and changes in its marginal cost. We information about managers' past identify some of the factors respon­ decisions. The fact that stock prices sible for this, and use an estimation have only an indirect role suggests procedure that sidesteps the most that the stock market may not be a

58. NBER Reporter Fall 1995 t Voluntary Export supply of skill in the economy in sion system, employment security . Restraints on Automobiles: the 1980s. We fmd that this is in­ laws, and centralized wage setting Evaluating a Strategic consistent with a simple explana­ in Sweden reinforced the distor­ Trade Policy tion in which supply shifts among tionary impact of the tax system. Steven Berry, women have played a major role. We describe the relevant Swedish James A. Levinsohn, and Instead, it supports the view that poliCies and institutional arrange­ Arielpakes demand shifts, rather than supply ments, and explain why the atten­ NBER Working Paper No. 5235 shifts, have been the underlying dant distortions are likely to have August 1995 cause of declining opportunities for hampered the efficient allocation of Industrial Organization, International less-skilled males and the rapid resources, reduced productivity, Trade and Investment, Productivity growth in inequality in the 1980s. and retarded economic growth and recovery. In May 1981, a voluntary export We also use state and SMSA-Iev­ We also develop evidence on restraint (VER) was placed on ex­ el data to estimate cross-substitu­ tion effects between men and the consequences of these distor­ ports of automobiles from Japan to women of different skill types. We tions for the size structure and in­ the United States. As trade policies find weak evidence that women dustrial distribution of employ­ go, this one was important. At may be substitutes for men who ment. Taking the U.S. industrial about the same time, although with are high school dropouts, and that distribution as a benchmark that much less fanfare, international college-educated women may have reflects a comparatively neutral set trade theorists were obtaining contributed to the growth in wage of policies and institutions, we find (then) startling results from models inequality by being better substi­ that Sweden's employment distri­ of international trade in imperfectly tutes for male high school dropouts bution is tilted sharply away from competitive markets. These models than for male high school gradu­ lower-wage industries, less capital­ suggested that in imperfectly com­ ates. We conclude with some sug­ intensive industries, and industries petitive markets, an activist trade gestive evidence that unmeasured characterized by greater employ­ ~ policy might enhance national wel­ shifts in demand that favored ment shares toward smaller firms fare. In this paper, we provide ~ skilled female workers over less­ and establishments. Compared to some empirical evidence on wheth­ skilled male workers may be bias­ other OECD economies, Sweden er these new theoretical possibilities ing our results toward fmding some has the lowest rate of self-employ­ might actually apply to the VERs. substitution between these two ment, a dominant role for larger groups. firms, and highly concentrated The Effects of Rising ownership and control of private­ Female Labor Supply sector economic activity. on Male Wages Industrial Policy, ChinhuiJuhn and Employer Size, and The Demand for DaeIlKim Economic Performance Illicit Drugs NBER Working Paper No. 5236 in Sweden Henry Saffer and August 1995 StevenJ. Davis and FrankJ.C~oupka JEL Nos. J31, J23 Magnus Henrekson NBER Working Paper No. 5238 Labor Studies NBER Working Paper No. 5237 August 1995 This paper examines the extent August 1995 Health Economics JEL Nos. L52, J21, H30 to which rapid increases in female This paper estimates the effects Labor Studies labor supply contributed to rising of heroin and cocaine prices and wage inequality and declining real The pre-1990 Swedish tax sys­ decriminalization of marijuana on wages for less-skilled males during tem strongly disfavored younger, the demand for those three drugs. the 1980s. We find that while the smaller, and less capital-intensive There are few prior empirical stud­ declines in the male wage are con­ firms and sectors, and discouraged ies in this area because data have centrated in the 1980s, growth in entrepreneurship and family own­ been difficult to acquire. We use the female labor supply slowed in ership of businesses in favor of in­ newly available data on drug prices the 1980s relative to the 1970s. stitutional ownership. Credit mar­ from the Drug Enforcement Agen­ ;. Women also increased their relative ket regulations, the national pen- cy, and are the first to link these

NBER RejJotterFalll995 59. data to a sample of 49,802 individ­ cymakers put more weight on sta­ and child outcomes may help us to. uals from the National Household ble employment than on stable in­ unra vel the complex process Survey of Drug Abuse. The results flation, therefore, their objective through which poverty is transmit- show that drug use is more respon­ can be attained more easily under ted across generations. sive to price than has been thought capital controls. previously. The participation price Capital Markets elasticity for heroin is about -.9 to Race, Children's Cognitive Integration, Volatility, -.8, and the participation price Achievement, and and Persistence elasticity for cocaine is about -.55 The Bell Curve Joshua Aizenman to -.36. Marijuana decriminalization Janet Currie and NBER Working Paper No. 5241 increases the probability of using Duncan Thomas August 1995 marijuana by about 4 to 6 percent. NBER Working Paper No. 5240 JEL Nos. F12, F21 The price elasticity for heroin is August 1995 International Trade and Investment about -1.8 to -1.6, and for cocaine JEL Nos. 120, 130 about -1.1 to -.72. We estimate Labor Studies This paper shows that volatility that legalization would lead to induces adverse first-order welfare about a 100 percent increase in the In The Bell Curoe, Herrnstein and effects in countries excluded from quantity of heroin consumed, and Murray demonstrate that a mother's the global capital market. We illus­ about a 50 percent increase in the score on the Armed Forces Qualifi­ trate this result in a model charac­ quantity of cocaine consumed. cation Test (AFQT) is a powerful terized by gains from a greater di­ predictor of her child's score on the vision of activities, in which shocks Can Capital Controls Peabody Picture Vocabulary Test are persistent. We show that non­ Alter the Inflation­ (PPVT). We replicate this finding linearities attributed to finandal au­ Unemployment Trade-Off? for PPVT and two related tests. tarky explain the adverse welfare However, even after controlling for Assaf Razin and Chi-Wa Yuen effects of volatility. We identify the AFQT, there are significant racial parameters determining the magni­ NBER Working Paper No. 5239 gaps in PPVT scores, suggesting August 1995 tude of the loss. It is proportional that AFQT is not all that matters. In to: the autocorrelation of shocks; JEL No. F2 fact, both maternal education and International Finance and volatility (as measured by the stan­ income are important determinants Macroeconomics dard deviation of shocks); and the of children's test scores, and their degree of product differentiation It is well known that in the influences differ dramatically with (as measured by the substitutability Mundell-Fleming model, capital the test, the child's age, and the among intermediate products). mobility creates a channel through child's race. These racial gaps in which permanent (transitory) tesl scores are important because, Trade, Spatial Separation, shocks to aggregate demand-such even within families, children with and the Environment as fiscal and trade shocks-are higher scores are less likely to re­ Brian R. Copeland and neutralized completely (partially) peat grades. Moreover, conditional M. Scott Taylor by the response of the real ex­ on children's test scores and mater­ NBER Working Paper No. 5242 change rate. An important policy nal AFQT, maternal education and August 1995 implication of the model that went income affect the probability of JEL Nos. FlO, Q20, RIO largely unnoticed is how cthe trans­ grade· repetition. We move beyond International Trade and Investment mission of these shocks under dif­ AFQT and examine the effects of ferent degrees of capital mobility individual Armed Services Voca­ We develop a simple two-sector may alter the inflation-unemploy­ tional Aptitude Battery subsets on dynamic model to examine the ef­ ment trade-off, that is, the Phillips children's scores. We find that the fects of international trade in the Curve. In the context of the sto­ skills that are rewarded in the labor presence of pollution-created cross­ chastic Mundell-Fleming model, market are not always the same as sectoral production externalities. we show that capital controls re­ those associated with improved We assume that the production of duce the output/employment varia­ child outcomes. An understanding "smokestack" manufactures gener­ tions at the expense of bigger vari­ of the relationship among different ates pollution, which lowers the ations in inflation rates. When poli- aspects of maternal achievement productivity of an environmentally

60. NBER Reporter Fall 1995 t sensitive sector ("fanning"). As a re­ hiring new personnel possessing the Bank Capital Regulation sult, the long-run production set is new technology, and incorporating in General Equilibrium nonconvex. Pollution provides a them into the existing structure. Gary Gorton and motive for trade, since trade can While the technological transfor­ Andrew Winton separate incompatible industries mation has been profound, bio­ NBER Working Paper No. 5244 spatially. Two identical, unregulat­ technology applications in this AUgust 1995 ed countries will gain from trade if large incumbent firm are more like­ JELNo. G21 the share of world income spent ly to be used in. combination with Corporate Finance on smokestack is high. In contrast, other technologies than in the new Using a multiperiod general when the share of world income biotechnology firms (NBFs), which equilibrium model of banking, we spent on the dirty good is low, tend to use biotechnology for both study whether the SOcially optimal trade can usher in a negatively re­ discovery and production of new level of stability in the banking sys­ inforcing process of environmental therapeutic entities. This difference tem can be implemented with reg­ degradation and real income loss in emphasis may result in value-en­ ulatory capital requirements. We for the exporter of smokestack. hancing synerg~es because of the show that: 1) bank capital is costly wealth of relat~d knowledge that because of the unique liquidity ser­ Present at the Revolution: makes for more effective applica­ vices provided by demand depos­ Transformation of tions of the new technologies. its, so a bank regulator optimally Technical Identity for However, it could also retard full may choose to have a risky bank­ a Large Incumbent adoption of biotechnology. Resolu­ ing system; and 2) even if the regu­ Pharmaceutical Firm After tion of this empirical question is lator prefers more capital in the the Biotechnological the subject of future research. system, the regulator is constrained Breakthrough It appears that this firm was by the private cost of bank capital, Lynne G. Zucker and somewhat slower than the dedicat­ which determines whether bank .. Michael R. Darby ed biotech firms to adopt the new shareholders will agree to meet J NBER Working Paper No. 5243 technology. Once the decision was capital requirements rather than August 1995 made to transform the technologi­ exit the industry. JELNo.031 cal identity of the firm, though, Productivity massive resources were provided The Causes and to recruit the intellectual human This paper presents a case study Consequences of Rate capital required to make it happen. Regulation in the Auto of the transformation in research We conducted a preliminary analy­ methods that occurred in a large Insurance Industry sis of data for this and other major Dwight M. Jaffee and U.S. pharmaceutical firm as a result pharmaceutical firms on the extent of the biotechnology revolution. Thomas Russell of publication by top "star" scien­ NBER Working Paper No. 5245 This transformation is not consis­ tists either affiliated with the firm or tent with the hypothesis that tech­ September 1995 writing with scientists from the firm JEL Nos. G22, G28 nological revolutions make existing -a powerful predictor of success­ firms obsolete (because of rigidity ful application of biotechnology. This paper examines various ex­ of core task routines). It is consis­ We also examined patenting of ge­ planations for the increase in the tent with the wealth-maximization netic sequences by these firms, as degree of regulation of the auto in­ hypothesis, under which valuable well as by corresponding NBFs. surance industry over the last ten assets (including delivery know­ The incumbent firms were general­ years. Using cross-sectional data how and other complementary ly slow to adopt biotechnology, but for California, we confirm earlier technologies) will not be wasted if as a group made great strides in fmdings for Massachusetts that the technological change in one part of the late 1980s in increasing their demand for auto insurance is high­ the organization is necessary for share of all commercial ties to the ly elastic to price. This implies that the remainder of the organization star scientists, as well as their share regulation-induced price rollbacks to maintain its competitiveness. of patents. (such as those mandated by Cali­ While the transformation was fornia's popular initiative, Proposi­ ~ achieved through a variety of tion 103) have significant effects on y methods, the primary route was welfare.

NBER RepotterFall 1995 61. We explain the increase in regu­ less obligated to its insureds than amount of franchising in a way that • lation in two ways: 1) as an attempt the mutual company. We assess the suggests that: 1) franchisors are not to lower rates to deal with the responses by stock and mutual able to control franchisees' prices problem of the uninsured motorist; firms to changes in their underwrit­ indirectly to the same extent that and 2) more fundamentally, as a ing environment from 1984 to 1991; they control company-owned unit response to the perceived lack of we use measures of individual prices; and 2) the prices in fran­ fairness of the sharp increase in firms' performance, by state and by chised and corporate units are sys­ premiums in the 1980s. This per­ insurance line, in eight different tematically different Finally, I show ception of lack of fairness arises lines of insurance. We find that that prices are systematically lower because, although auto insurance stock companies are more likely in corporate restaurants. This sug- costs rose sharply in the 1980s, than mutuals to reduce their busi­ gests that the reason behind the most buyers of auto insurance ness in unprofitable situations, and price differentials is not franchisor have no claims in any ten-year pe­ they have higher losses than mutuals opportunism, but more likely dou- riod. Thus, most buyers have only for a given amount of premiums. ble marginalization or, potentially, last year's premium as a reference the existence of positive horizontal point with which to judge the fair­ Pricing Decisions in externalities among restaurants in a ness of this year's premium. Franchised Chains: A Look chain. We test the hypothesis that the at the Restaurant and increase in regulation is driven by Fast-Food Industry Economic Issues in Vaccine a perception of unfairness by ana­ Francine Lafontaine Purchase Arrangements lyzing the cross-county voting pat­ NBER Working Paper No. 5247 David S. Salkever and tern on Proposition 103. Voting in September 1995 Richard G. Frank favor of price regulation is correlat­ JEL Nos. L22, L42, L8 NBER Wolking Paper No. 5248 ed positively with the level of in­ Industrial Organization September 1995 surance premium. This result is This paper examines empirical JEL No. 11 consistent both with the view that issues of pricing and price disper­ Health Care. Health Economics, voting behavior is based on self­ sion within franchised restaurant Industrial Organization interest and with the view that the and fast-food chains. Given the per Federal purchases of major child­ increased demand for regulation is se illegality of resale price mainte­ hood vaccines account for roughly driven by concerns that the large nance under current u.s. antitrust half of the total market for these disparity in premiums across coun­ laws, and the fact that franchised vaccines. This paper examines fed­ ties is unfair. outlets are independent businesses eral purchasing practices in the under the law, franchisors must context of the recent literature on Organizational Form and delegate the power to set prices to bidding and procurement, and com­ Insurance Company franchisees, whereas corporate pares these practices to UNICEF Performance:Stodks chains can control downstream pri­ vaccine procurement arrangements. Versus Mutuals ces directly. I examine whether it Federal contracts were awarded to Patricia Born, matters empirically who, either the a single winner, and the firms eligi­ William M. Gentry, franchisor or the franchisee, gets to ble to bid were limited in number W. Kip Viscusi, and choose downstream prices, and if Richard]. Zeckhauser (since the number of u.s. licensed so, why. firms is small). Since production NBER Working Paper No. 52"46 After discussing a number of capacity cannot be expanded September 1995 reasons why prices chosen by fran­ quickly, and the federal share of Corporate Finance, chisees may differ from those that purchases is large, we hypothesize Industrial Organization a franchisor would pick, I use data that firms' bid prices will be higher One unusual feature of the u.s. from all restaurant chains in the for larger contracts. This paper property-casualty insurance indus­ metropolitan Pittsburgh and Detroit analyzes contracts over 1977-1992 try is the coexistence of stock and areas to show that there is price to determine the relationship be­ mutual companies. Agency theories dispersion in fast-food franchising. tween contract size and price, as suggest that the stock company I then show that the amount of well as the effects on contract pri­ may be more "opportunistic" and price dispersion relates to the ces of the National Vaccine Injury

62. NBER Reporter Fall 1995 \

t Compensation Pro~ram enacted in duce the amount of trade and have resources are significantly more .. 1987 and changes m the number of adverse consequences on welfare. likely to participate. This finding firms in the market. Our results Conversely, policies that subsidize differs from widespread beliefs that provide equivocal support for a assimilation and the acquisition of eligible individuals are discouraged positive relationship between size majority language skills can be so­ by the difficulty of the application and price and some evidence of a cially beneficial. process, or that many are unin­ positive effect on price of the inju­ I test and confirm the theory by formed about the program. ry compensation program. examining U.S. Census data. It re­ veals that the likelihood that an im­ Optimal Inflation Targets, Culture and Language migrant will learn English is related "Conservative" Central Edward P. Lazear inversely to the proportion of the Banks, and Linear NBER Working Paper No. 5249 local population that speaks his or Inflation Contracts September 1995 her native language. Lars E. o. Svensson JEL Nos. D3, JOO, J15, J18 NBER Working Paper No. 5251 Labor Studies Factors Det~mining September 1995 Participation of the JEL Nos. 120, 130 Common culture and common Elderly in SSI Labor Studies language facilitate trade between Kathleen McGarry individuals. Minorities have incen­ Inflation-target regimes (includ­ NBER Working Paper No. 5250 tives to become assimilated and to ing those of New Zealand, Canada, learn the majority language so that September 1995 the United Kingdom, Sweden, and they have a larger pool of potential JEL No. 13 Finland) are interpreted as having Aging trading partners. The value of as­ explicit inflation targets and implic­ similation is larger to an individual The same low participation rates it output/unemployment targets. from a small minority than to one that plague many welfare programs Without persistence in output/un­ ..a from a large minority group. When have been observed among the el­ employment, delegating monetary • a society has a very large majority derly who are eligible for Supple­ policy to an independent central of individuals from one culture, in­ mental Security Income (SS!). A bank with a discretionary instru­ dividuals from minority groups will number of hypotheses have been ment and an optimal inflation tar­ be assimilated more quickly. As­ offered to explain the low enroll­ get can: eliminate the discretionary similation is less likely when the ment, but none has attracted uni­ inflation bias; mimic the optimal immigrants' native culture and lan­ versal acceptance. In this paper I linear inflation contract suggested guage are represented broadly in use the Survey of Income and Pro­ by Walsh, and extended by Pers­ their new country. Also, when gov­ gram Participation to examine the son and Tabellini; and achieve the ernments protect minority interests participation of the elderly in SS!. equilibrium corresponding to an directly, incentives to be assimilat­ Because of the high quality of the optimal rule with commitment. ed into the majority culture are re­ data, I am able to determine eligi­ Thus an "inflation-target-conserva­ duced. Both factors may explain bility more accurately than in most tive" central bank, with an inflation the recent rise in multiculturalism. previous studies. In this sample, target equal to the socially best in­ Individuals do not internalize the only 56 percent of those whom I flation rate minus any inflation social value of assimilation prop­ determine to be eligible for SSI bias, dominates a Rogoff "weight­ erly. They ignore the benefits that presently are receiving benefits. I conservative" central bank with in­ others receive when they learn the model the decision to participate as creased weight on inflation stabi­ majority language and become as­ a probit equation, but modify the lization, which if?creases the vari­ similated. In a pluralistic society, a likelihood function to account for ability of output/unemployment in government policy that encourages measurement error in the expected a suboptimal way. diverse cultural immigration over benefit. The results indicate that With persistence in output/un­ concentrated immigration is likely participation is determined primari­ employment, a constant inflation to increase the welfare of the na­ ly by the financial situation of the target is equivalent to a constant tive population. In the absence of eligible individuals. Although all linear inflation contract. They both ~ strong offsetting effects, policies those eligible for SSI are poor, can eliminate the average inflation y that encourage multiculturalism re- those with little in the way of other bias, but not the state-contingent

NBER RejJOrterFall 1995 63. portion of the inflation bias. Infla­ nancial asset holdings by 1.4 to 5.6 to be weak price substitutes at tion variability is too high, and out­ percent, so that UI crowds out up best, and in fact may be price com­ put variability too low, compared to half of private savings for the plements. Taken together, these to the equilibrium corresponding typical unemployment spell. We fIDdingS indicate that multinational to an optimal rule. An optimal also fmd that this effect is stronger outsourdng contributed very little state-contingent inflation target can for those facing higher unemploy­ to rising wage inequality. remove all inflation bias; but, in ment risk and weaker for older contrast to an optimal state-contin­ workers. Labor Income Indices gent linear inflation contract, it still Designed for Use in leaves inflation variability too high. Multinational Contracts Promoting Delegation with an optimal state­ Corporations, Outsourcing, Income Risk Management contingent inflation target to a Ro­ and American Wage RobertJ.S~erand goff "weight-conservative" central Divergence Ryan Schneider bank then can achieve the equilib­ Matthew J. Slaughter NBER Working Paper No. 5254 rium corresponding to an optimal NBER Working Paper No. 5253 September 1995 rule. Inflation targets on average September 1995 JEL No. J31 may be exceeded, and they may JEL Nos. F21, F23, J31 Economic Fluctuations have imperfect credibility. Never­ International Trade and Investment, theless they may reduce inflation Labor Studies In this paper, we create indexes usefully, and they appear much of labor income for groups of indi­ easier to implement than linear in­ Many economists studying Amer­ viduals, using data from the Panel flation contracts. ica's wage divergence in the 1980s Study of Income Dynamics. People have concluded that its primary are grouped according to jobs, cause was a shift within industries UnemplQyment Insurance education level, and skill category. in relative labor demand toward The groups are defIDed so that rel­ and Precautionary Saving those with more skills. FollOwing Eric M. Engen and atively few people move between the modeling framework and em­ them. For each of the groupings, Jonathan Gruber pirical methods developed in NBER Working Paper No. 5252 we generate a labor income index; Slaughter (1993), I try in this paper we then describe similarities be­ September 1995 • to determine the extent to which tween pairs of indexes and be­ JEL Nos. J65, H53, E21 outsourcing by multinational cor­ Labor Studies, Public Economics tween indexes and individual labor porations contributed to this shift incomes. We argue that indexes We consider both theoretically in labor demand. I use data from like those presented here someday and empirically the effect of unem­ the Bureau of Economic Analysis might be used in settlement formu­ ployment insurance (UI) on pre­ on U.S. manufacturing multination­ las in contracts promoting income cautionary savings behavior. Simu­ als in the 1980s. My main fIDding is risk management. lations of a stochastic life-cycle that the data are inconsistent with model suggest that increasing the U.S. multinationals having out­ Trade and Production generosity of UI will lower the as­ sourced heavily in the 1980s. Networks of U.S. MNCs set holdings of the median worker First, I construct a set of stylized and Exports by Their substantially, and that this effect facts about the employment, in­ Asian Afiuiates vestment, and production patterns will both rise with unemployment Robert E. Lipsey risk and fall with worker age. We of these firms. I find that most of NBER Working Paper No. 5255 test these implications by matching these facts are inconsistent with September 1995 data on potential UI replacement widespread outsourcing. Second, JEL Nos. F21, F23 rates to asset holdings in the Sur­ to test more rigorously whether International Trade and Investment vey of Income and Program Partici­ these firms substitute between U.S. pation. Our empirical results are and foreign production labor, I es­ Network connections within quite consistent with the predic­ timate their factor-price elastidties multinational corporations (MNCs) tions of the model. We find that of demand in a translog-cost-func­ seem to improve export market raising the replacement rate for UI tion specification. I find that home shares for their Asian affiliates. In by 10 percentage points lowers fi- and foreign production labor seem particular, Asian affiliates of U.S.

64. NBER Reporter Fall 1995 ·t MNCs export more to markets derwriting were combined closely suggest that it has a positive impact .. where their parent firms' exports t? in the departmental structure. We on a child's cognitive achievement. affiliates are larger, and less to find that bank managers were con­ markets where their parent firms' cerned about such perceptions What Have exports go more to nonaffiliates. during this period. Macroeconomists Learned However, the latter effect is much We then develop further tests to About Business Cycles smaller per dollar of parent exports support the view that by distancing from the Study of than the former effect. These rela­ underwriting activities from lending Seasonal Cycles? tionships are fairly consistent operations, banks could certify the Jeffrey A. Miron and across industries and markets, quality of the issues they under­ J. Joseph Beaulieu across markets within two indus­ wrote more credibly, thereby ob­ NBER Working Paper No. 5258 tries, across industries for two affili­ taining higher prices for them. Our September 1995 ate home countries, and across ex­ results suggest that internal organi­ JEL Nos. E30, E32 porters and industries for individ­ zation indeed may affect the activi­ Economic Fluctuations, ual markets. ties and efficiency of a firm. They Monetary Economics also suggest th~t bank regulators' Organization Structure interest in "firewalls" between com­ This paper argues that analysis and Credibility: Evidence mercial and investment banking of seasonal fluctuations can shed from Commercial Bank may be reasonable, but that the light on the nature of business cy­ Securities Activities Before market may propel banks to adopt cle fluctuations. The fundamental the Glass-Steagall Act an internal structure that would ad­ reason is that in many instances, RandallS.Krosznerand dress regulators' concerns. identifying restrictions about sea­ Raghuram G. Rajan sonal fluctuations are more believ­ NBER Working Paper No. 5256 The Impact of Child able than analogous restrictions September 1995 Health and Family about nonseasonal fluctuations. We Corporate Finance Inputs on Child show that seasonal fluctuations Cognitiye Development provide good examples of shifts in This paper investigates how or­ preference and synergistic equilibri­ ganizational structure can affect a Robert Kaestner and Hope Corman ums. We also find evidence against firm's ability to compete. In partic­ production smoothing and in favor NBER Working Paper No. 5257 ular, we examine the two ways in of unmeasured variation in labor September 1995 which U.S. commercial banks orga­ and capital utilization. In some in­ Health Economics nized their investment banking op­ dustries capacity constraints appear erations before the 1933 Glass­ In this paper we analyze the im­ to bind. Steagall Act forced them to leave pact of child health and other fami­ the securities business: as an inter­ ly characteristics on the cognitive Globalization, nal securities department within the achievement of children between Convergence, and History bank, and as a separately incorpo­ the ages of five and nine. We esti­ Jeffrey G. Williamson rated and capitalized securities affil­ mate both cross-sectional and NBER Wolking Paper No. 5259 iate. We document a strong move­ fixed-effects models using data September 1995 ment toward the use of the affiliate from the National Longitudinal Sur­ Development of the American Economy structure during the 1920s; regula­ vey of Youth. Several of our results tion does not appear to explain this challenge the conclusions found in There were three epochs of evolution. While departments un­ the existing literature. First, we find growth experien~e after the mid- derwrote seemingly higher-quality only a weak relationship between 19th century for what is now called firms and securities than compara­ several measures of child health the "OECD club": the late-19th cen­ ble affiliates did, the departments and cognitive development. Sec­ tury; the middle years between obtained lower prices for the issues ond, we find that additional mater­ 1914 and 1950; and the late-20th they underwrote. This is consistent nal schooling does not improve the century. The late-19th and the late- with the hypothesis that there was child's cognitive achievement. Fi­ 20th century epochs saw overall ;l a perception of potential conflicts nally, our estimates of the effect of fast growth and convergence: poor y of interest when lending and un- mother's labor force participation countries tended to grow even fast-

NBER RepotterFall 1995 65. er than rich ones, and the econom­ taken in a stochastic, dynamic envi­ We sununarize a dynamic mOdel. ic gap between rich and poor ronment that allows us to charac­ of the demographic structure of Ja- countries diminished. The middle terize the relationship between pan. It is capable of tracing the dy­ years saw overall slow growth and lumpy investment and the state of namic development of the Japa- divergence: poor countries tended the aggregate economy. We sup­ nese population, including the dis­ to grow even slower than rich plement our theoretical results with tribution of families by age, sex, ones, and the economic gap be­ numerical and empirical analyses and marital status of the head, as tween rich and poor countries wi­ of the dynamics of lumpy invest­ well as by the number and age of dened. Since the middle years also ment at the plant level and the as­ children and other dependents. We involved economic autarky and sociated aggregate implications. combine this model with a specifi- "deglobalization," while the other The dynamics are surprisingly rich, cation of the processes generating periods saw increasing globaliza­ since they represent the interaction family income and consumption, tion in world conunodity and fac­ among a replacement cycle, the and use it to generate the pattern tor markets, history offers an un­ cross-sectional distribution of the of aggregate income, saving, and ambiguous positive correlation be­ age of the capital stock, and the asset accumulation for 1985-2090 tween globalization and conver­ state of the aggregate economy. under alternative assumptions gence. But is that correlation spuri­ Our empirical analysis of these about fertility. The results suggest ous? When I examine the pre­ dynamics is based on plant-level . that the saving-income ratio for Ja- World War I years in detail, I find investment data for the Longitudi­ pan will increase slightly in the im­ that the correlation is causal: the nal Research Database for 1972-91. mediate future as the number of globalization of commodity and Overall, we find that the frequency children per family declines sharp­ factor markets served to playa crit­ of lumpy investment activity is ly, and then will fall moderately as ical, perhaps the critical, role in higher during periods of high eco­ the proportion of older persons in contributing to convergence. A nomic activity, and is more likely the population increases. Quantita­ century and a half of OEeD club as capital ages. These results are tive results depend critically on the history also suggests that econo­ consistent with the predictions of labor force participation rate of mists should pay more attention to our theoretical model. Nonetheless, older persons and on the probabili­ who gains and who loses from the predicted path of aggregate in­ ty of older persons merging into convergence, since the answers vestment that neglects the interac­ younger households. However, un­ may help to determine whether tion of the nonflat hazard and the less some major changes in Japa­ proglobalization or antiglobaliza­ cross-sectional distribution of the nese saving behavior take place, tion policies will persist. age of the capital stock tracks actu­ our analysis suggests that Japan al aggregate investment quite well. will have an unusually high net Machine Replacement However, ignoring the fluctuations worth-income ratio as its popula­ and the Business Cycle: in the cross-sectional distribution tion stabilizes or begins to decline. Lumps and Bumps can yield predictable nontrivial er­ Russell Cooper, rors in forecasting changes in ag­ Asset Pricing Lessons for John C. Haltiwanger, and gregate investment during periods Modeling Business Cycles Laura Power following large swings in aggregate Michele Boldrin, NBER Working Paper No. 5260 investment. LawrenceJ.Ch~tiano,and September 1995 Jonas Fisher Economic Fluctuations Demographic Dynamics, NBER Working Paper No. 5262 September 1995 This paper explores cyclical fluc­ Labor Force Participation, JEL Nos. 131, 023 tuations in investment attributable and Household Asset Asset Pricing, Economic Fluctuations to discrete changes in the plant's Accumulation: Case stock of capital. We focus on a ma­ of Japan We develop a model that ex­ chine replacement problem in Albert Ando and Andrea Moro plains the observed equity premi­ which a producer decides whether NBER Working Paper No. 5261 um and average risk-free rate, with­ to replace its entire existing stock September 1995 out implying counterfactually high of capital with new machinery and JEL Nos. D12, E21, J10 risk aversion. The model also does equipment. This decision is under- Economic Fluctuations well in accounting for business cy-

66. NBER Reporter Fall 1995 cle phenomena. With respect to the An Empirical Analysis of from Tiebout's vision of the local t conventional measures of business the Welfare Magnet Debate public goods producer as an entre- cycle volatility and comovement Using the NLSY preneur, which is unrealistic be- with output, the model does roug- Phillip B. Levine and cause local public goods are non- hly as well as the standard business DavidJ. Zimmerman verifiable. The Tiebout mechanism cycle model On two other dimen- NBER Working Paper No. 5264 does not operate in alternative sions, the model's business cycle September 1995 models of the local public goods implications actually are improved. Labor Studies producer, including the bureaucra- Because of enhanced internal prop- cy and agenda models. None of agation, the model explains the fact This paper examines the extent these models is useful for predict- that there is positive persistence in to which differences in the gen- ing how producers of local public output growth; the model also pro- erosity of welfare across states goods react to policies, such as vi des a resolution to the "excess leads to interstate migration. Using equalization of school spending, sensitivity puzzle" for consumption microdata from the National Longi- that change the structure of local and income. The key features· of tudinal Survey of Youth (NLSY) be- public fmance. the model are habit persistence tween 1979 and i992, we employ a This paper builds a theory of preferences and a multisector tech- quasi-experimen al design that uses the producer that draws upon Tie- nology with limited intersectoral the categoricar eligibility of the bout's mechanism and the theory mobility of factors of production. welfare syste11J-. We compare the of incentives for regulation. The es- pattern of cros~-state moves among sential insight is that Tiebout's What Do Budget poor single women with children mechanism generates information Deficits Do? who are likely to be eligible for that can be used in regulatory Laurence M. Ball and benefits to the pattern observed schemes to achieve lower costs for N. Gregory Mankiw among other p~or households. We any given provision of local public NBER Working Paper No. 5263 find little evidence that welfare-in- goods. Thus, we face a fundamen- September 1995 duced migration is a widespread tal trade-off between promoting t Economic Fluctuations, phenomenon. equitable consumption of the pub- Monetary Economics lic good and promoting efficiency Is There an Equity- (cost minimization) in the produc- This paper discusses in four Efficiency Trade-Off in tion of the public good. This trade- steps the effects of budget deficits School Finance? Tiebout off exists even when equity in con- on the economy. First, it reviews and a Theory of the Local sumption generates positive exter- standard theory about how budget Public Goods Producer nalities, as is often suggested of the deficits influence saving, invest- Caroline M. Hoxby consumption of schooling. I show ment, the trade balance, interest that when the Tiebout mechanism NBER Working Paper No. 5265 rates, exchange rates, and long- for schools is weakened by state- September 1995 term growth. Second, it offers a level school funding, per-pupil Public Economics rough estimate of the magnitude of costs rise and the growth of educa- some of these effects. Third, it dis- New empirical work shows that tional attainment falls. This latter fact cusses how budget deficits affect the degree of competition among implies that losses from inefficient economic welfare. Finally, it con- public providers of local public production generally outweigh siders the possibility that continu- goods, or between public and pri- gains from equalized consumption. ing budget deficits in a country vale providers of local public could lead to a "hard landing" in goods, matters. This evidence needs Forbidden Payment: which the demand for the coun- a theory of the local public goods Foreign Bribery and try's assets suddenly collapses. producer. Tiebout's hypothesis American Business spawned a literature that gives us a After 1977 useful theory of the consumer, and James R. Hines,Jr. the same mechanism can generate NBER Working Paper No. 5266 a theory of the local producer of September 1995 public goods. This potential has re- JEL Nos. F23, H87, 073 , mained largely undeveloped, apart Public Economics

NBER ReponerFalll995 67. The United States prohibits Amer­ the entire sample period, for sub­ the commission of that crime, in • ican individuals and corporations periods, and for individual years. which deterrence and incapacita- from bribing foreign government These results indicate that a higher tion are indistinguishable, it can be officials. Legislation enacted in sensitivity cannot be interpreted as demonstrated that these two forces 1976 and 1977 stipulates tax penal­ evidence that a firm is more fman­ act in opposite directions when ties, fines, and even prison terms cially constrained. We discuss how looking across a variety of crimes. for executives of American compa­ and why the opposite may be true. Incapacitation suggests that an in­ nies that pay illegal bribes. This pa­ These findings challenge much of crease in the arrest rate for one per examines the effect of that leg­ the existing evidence on the effects crime will reduce all crime rates; islation on the operations of U.S. of fmancial constraints. deterrence predicts that an increase firms in bribe-prone countries after in the arrest rate for one crime will 1977. Four separate indicators re­ Why Do Increased Arrest lead to a rise in other crimes as veal that U.S. business activities in Rates Appear to Reduce criminals substitute away from the those countries fell sharply after Crime: Deterrence, first crime. Empirically, deterrence passage of the Foreign Corrupt Incapacitation, or appears to be the more important Practices Act of 1977. These results Measurement Error? factor, particularly for property suggest that this unilateral action by Steven D. Levitt crimes. the United States served to weaken NBER Working Paper No. 5268 the competitive positions of Ameri­ September 1995 Inside Money, Outside can flIffiS without significantly reduc­ JEL No. K42 Money, and Short-Term ing the importance of bribery to Public Economics Interest Rates foreign business transactions. A strong, negative empirical cor­ v.v.chari, relation exists between arrest rates LawTenceJ.~tiano,and Martin Eichenbaum. Do Financing Constraints and reported crime rates. While this Explain Why Investment relationship often has been inter­ NBER Working Paper No. 5269 • Is Correlated with preted as support for the deter­ September 1995 Cash Flow? rence hypothesis, it is equally con­ JEL Nos. E3, E4, E5 Steven N. Kaplan and sistent with incapacitation effects, Economic Fluctuations, Luigi Zingales and/or a spurious correlation that Monetary Economics NBER Working Paper No. 5267 would be induced by measurement This paper presents a quantita­ September 1995 error in reported crime rates. This tive general equilibrium model Corporate Finance, Public Economics paper attempts to discriminate with multiple monetary aggregates. This paper investigates the sour­ among deterrence, incapacitation, The framework incorporates a ces of the correlation between cor­ and measurement error as explana­ banking sector and distinguishes porate cash flow and investment. tions for the empirical relationship among M1, the monetary base, cur­ We undertake an in-depth analysis between arrest rates and crime. rency, and various measures of re­ of the 49 low-dividend ftrms identi­ Using a modified version of the serves: total, excess, and nonbor­ fied by Fazzari, Hubbard, and Pe­ techniques of Griliches and Haus­ rowed. We use a variant of the tersen (988) as having an unusual­ man (986) for dealing with mea­ model to analyze two sets of em­ ly high sensitivity of investment to surement error in panel data, this pirical facts: first, different mone­ cash flow. We fmd that in only 15 paper· first demonstrates that the tary aggregates covary differently percent of firm-years is there some presence of measurement error with short-term nominal interest question as to a firm's ability to ac­ does not appear to explain the ob­ rates. Broad monetary aggregates cess internal or external funds to served relationship between arrest such as M1 and the monetary base increase investment. Strikingly, rates and crime rates. To differenti­ covary pOSitively with current and those firms that appear less finan­ ate between deterrence and inca­ future values of short-term interest cially constrained exhibit a signifi­ pacitation, I examine the impact of rates, while the nonborrowed re­ cantly greater sensitivity of invest­ changes in the arrest rate for one serves of banks covary negatively ment to cash flow than firms that crime on the rate of other crimes. with current and future interest appear more financially con­ In contrast to the effect of in­ rates. Observations such as this strained. We find this pattern for creased arrests for one crime on "sign switch" are at the core of re-

68. NBER Reporter Fall 1995 cent debates about the effects of State and Local nental trade blocs, an open region­ monetary policy actions on short­ Pension Plans alism in which trade blocs under­ term interest rates. According to Olivia S. Mitchell and take relatively modest external lib­ our model, the sign switch occurs Roderick Carr eralization usually can produce Pa­ because movements in nonbor­ NBER Working Paper No. 5271 reto improvement. In the bilateral rowed reserves are dominated by September 1995 trade data for 1970-92, there are exogenous shocks to monetary Aging, Labor Studies indeed regions that, while exhibit­ policy, while movements in the ing an inward trade bias, neverthe­ base and M1 are dominated by en­ This paper examines the role less are consistent with this notion dogenous responses to nonpolicy and function of pension plans cov­ of open regionalism. shocks. Second, broad monetary ering state and local government aggregates covary positively with employees in the United States. The Effects of Low output. We quantify the Friedman Covering about 16 million employ­ Birthweight and Other and Schwartz hypothesis that this ees (including teachers, fire fight­ Medical Risk Factors on covariation reflects the effects of ers, police, members of the judicia­ Resource Utilization in exogenous shocks to monetary ry, and many oilier state and local the Preschool Years policy, and the hypothesis that employees), these plans manage a Hope Corman they reflect the endogenous re­ substantial stock of fmancial assets NBER Working Paper No. 5273 sponse of monetary aggregates to --close to $1 trillion-and receive September 1995 shocks in the private economy. annual contributions from employ­ Health Economics ees and government revenues to­ More Bad News for taling about $56 billion. Using data This study compares the utiliza­ Smokers: The Effects of gathered from a variety of different tion of resources of preschool-aged Cigarette Smoking on sources, some of which only re­ children who are at medical risk Labor Market Outcomes cently have become available, we with that of their healthier pre­ Phillip B. Levine, describe the benefits, financing, school-aged peers. Medical risk is Tara A. Gustafson, and and management of these plans, defined as having been born at Ann D. Velenchik and identify some of the prominent low birthweight, having an activity NBER Working Paper No. 5270 challenges facing them in the next limitation, or having a chronic September 1995 decade. health or handicapping condition. Health Economics, Labor Studies The resources include: child care, preschool, kindergarten, Head Start This paper uses data from the Open Regionalism in a programs, and medical resources. National Longitudinal Survey of World of Continental Youth to examine the effect of Trade Blocs The study uses two distinct data­ sets: the National Health Interview smoking on wages and employ­ Shang-Jin Wei and ment. The panel nature and house­ Jeffrey A. Frankel Survey's Child Health Supplement of 1988, with approximately 2500 hold structure of these data enable NBER Working Paper No. 5272 children aged 3 to 5; and the Na­ us to implement methods to ac­ September 1995 tional Household Education Survey count for differences in observed International Trade and Investment and unobserved individual charac­ of 1991, which consisted of about teristics that may be correlated with Continental trade blocs are 6700 children who were aged 3 to both smoking and wages. We esti­ emerging in many parts of the 5. To explore differences between mate changes in wages assodated world almost in tandem. If trade at-risk and healthier peers, I hold with changes in smoking behavior blocs are required to satisfy the constant a variety of social and and models that use sibling com­ McMillan criterion of not lowering economic factors. parisons to address the potential their trade volume with outside I find consistently that at-risk problem of heterogeneity. Esti­ countries, then they have to en­ preschool-aged children are more mates from alternative spedfications gage in a dramatic reduction of likely to become hospitalized and all indicate that smoking reduces trade barriers against nonmember are less healthy than their peers wages by roughly 4 to 8 percent. countries. That may not be politi­ who are not at risk. In addition, ~ We observe no robust, statistically cally feasible. On the other hand, they are more likely to delay entry , significant effect on employment. in a world of simultaneous conti- into kindergarten. There is no evi-

NBER Reporter Fall 1995 69. dence for differences in the Rate Regulation and the This paper examines the rela­ amount or type of child care, or in Industrial Organization of tionship among openness, trade, mothers' labor force participation. Automobile Insurance and migration in the Asia-Pacific There is some evidence that the at­ Sharon Tennyson and region after 1970. Conventional risk children consume more pre­ SusanJ. Suponcic reduced-form empirical-growth spe­ school resources. NBER Working Paper No. 5275 cifications are augmented by an September 1995 appeal to structural modeling, an Does Measured School JEL Nos. G22, L51, L11 extension that reveals a rich set of Quality Really Matter? interactions among policy, distor­ An Examination of the This paper analyzes the impact tions, factor accumulation, and Earnings-Quality of rate regulation on the structure growth. A broad array of openness Relationship of insurance markets for private measures playa major role in the passenger automobile insurance. JamesJ. Heckman, successful growth performance of Anne Layne-Farrar, and We argue that states' restrictions on the Asia-Pacific region, a key chan­ Petra Todd automobile insurers' rates of return nel being the distortion-investment will distort the structure of the mar­ NBER Wolking Paper No. 5274 nexus. In contrast, the results sug­ ket by distorting insurers' entry and September 1995 gest little role for migration as a output decisions. Cross-sectional Labor Studies quantitatively Significant growth analysis of the numbers of firms determinant, at least at the macro This paper examines the eco­ and the relative market shares of level; this is no surprise in this area nomic and empirical foundations firms of different organizational of historically low net rates of mi­ of the aggregate evidence on the characteristics supports this argu­ gration. However, I find that with­ effect of schooling quality on earn­ ment, especially for those states in-sample prediction for the ings. We present a common frame­ that impose the most stringent reg­ Asia-Pacific region is harder to work that nests all previous studies ulation. The analysis suggests that achieve: "good luck" as well as as special cases. We discuss two increased regulatory stringency "good policy" played a part. crucial identifying assumptions and lowers the total number of firms test them. The first assumption is selling in the market, and lowers Employee Ownership, the absence of interactions be­ the number of low-cost and nation­ tween region of birth and region of Employee Attitudes, and al firms in the market. The market Firm Performance residence in the return to school­ shares of these latter two groups of Douglas Kruse and ing. This rules out patterns of mi­ firms also are lowered significantly Joseph Blasi gration on the basis of realized by increased regulatory stringency. NBER Working Paper No. 5277 earnings ih the destination state; These findings hold even after we September 1995 using 1970, 1980, and 1990 Census control for other factors that may Labor Studies data, we decisively rejected this hy­ influence the relative prevalence of pothesis. A second assumption is these firms in the market, and they Employee ownership in U.S. that log-earnings-equations are lin­ are robust to the assumption that companies has grown substantially ear, or nearly linear, in schooling. regulatory stringency itself in a in the past 20 years. This paper re~ This assumption is false. We find state is determined partially by the views and provides some meta­ that estimated earnings-quality re­ number and market shares of large, analyses on the accumulated evi­ lationships are sensitive to specifI­ low-cost producers. dence concerning the prevalence, cation of the earnings function. causes, and effects of employee When false linearity assumptions Growth and Convergence ownership; it covers 25 studies of are relaxed, the only effect of mea­ in the Asia-Pacific Region: employee attitudes and behaviors, sured schooling quality is on the On the Role of Openness, and 27 studies of productivity and returns for college graduates. The Trade, and Migration profitability (with both cross-sec­ evidence for an aggregate earn­ Alan M. Taylor tional and pre/post comparisons). ings-quality relationship is weak NBER Working Paper No. 5276 Attitudinal and behavioral stud­ once false empirical restrictions are September 1995 ies tend to find higher employee relaxed. JEL Nos. E22, F43, 040 commitment among employee­ Development of the American Economy owners but mixed results on satis-

70. NBER Reporter Fall 1995 faction, motivation, and other mea­ propriability into the technical The apparent success of several sures. Perceived participation in change literature, by adding suppli­ East Asian countries in sterilizing decisions in itself is not increased er relations as a determinant of capital inflows seems to contradict automatically through employee technology adoption. Second, the findings of high capital mobility. ownership, but may interact posi­ pa per extends transaction-cost This paper argues that empirical tively with employee ownership in analysis, by relaxing the assump­ studies examining money market affecting attitudes. tion that agents' private maximizing rates may be misleading, since While few studies individually behavior always will produce orga­ most lending is mediated through find clear links between employee nizational forms that maximize domestic banking systems. In de­ social effiCiency. ownership and firm performance, veloping countries with repressed meta-analyses favor an overall pos­ domestic financial markets, yields itive association with performance The Term Structure of on bank deposits might be tied for employee stock ownership Interest Rates and Its Role closely to international interest plans and for several cooperative in Monetary Policy for the rates, but bank loan rates might be features. The dispersed results European Central Bank more independent. We use a sim­ among attitudinal and performance Arturo Estrella and ple open-economy macro model studies indicate the importance of Frederic S. MiShkin incorporating bank credit to moti­ firm-level employee relations, hu­ NBER Working Paper No. 5279 vate alternative tests of financial man resource policies, and other September 1995 market integration. We find that circumstances. JEL Nos. E5, E17 capital inflows affect bank lending Monetary Economics in cases in which deposit and loan Supplier Relations and This paper examines the rela­ markets are integrated with world Adoption of New tionship of the term structure of in­ markets, and hence in which steri­ Technology: Results of terest rates to monetary policy in­ lization is not effective. In cases in Survey Research in the struments and to subsequent real which loan rates are more inde­ U.S. Auto Industry activity and inflation in both Eu­ pendent, sterilization seems to be Susan Helper rope and the United States. The re­ more effective. Next, we examine NBER Wolking Paper No. 5278 sults show that monetary policy is the effect of bank lending on eco­ an important determinant of the September 1995 nomic activity. The data suggest Productivity term structure spread, but it is un­ that the link between bank credit likely to be the only determinant. and investment is important in Using an original data source, In addition, there is significant pre­ countries with isolated bank loan this paper investigates the circum­ dictive power for both real activity markets. stances under which firms adopt and inflation. The yield curve is computer numerical control (CNC), thus a simple and accurate mea­ Would Privatizing Social an important type of flexible auto­ sure that should be viewed as one Security Raise mation that can increase signifi­ piece of useful information that, Economic Welfare? cantly productivity, product variety, along with other information, can Martin Feldstein and quality. The paper shows that be used to help guide European NBER Wolking Paper No. 5281 arms'-length supplier/customer re­ monetary policy. September 1995 lationships are a significant barrier JEL No. H55 to CNC adoption, even when CNC Asia-Pacific Capital would improve efficiency. For Markets: Integration Public Economics firms in which CNC would be effi­ and Implications for A funded social security retire­ cient, but that currently receive lit­ Economic Activity ment program would imply a larg­ tle commitment from their custo­ Menzie Chinn and er capital stock and a higher level mers, an increase in contract length Michael Dooley of real income than an unfunded of one year would increase the NBER Working Paper No. 5280 program that provides the same adoption rate by 30 percent. September 1995 level of benefits. The transition These results have theoretical JEL Nos. F36, F41 from an unfunded program to a ~ implications in two areas. First, the International Finance and funded program that does not re­ , paper integrates questions of ap- Macroeconomics duce the benefits of existing retir-

NBER Reporter Fall 1995 71. ees or the present value of the both the target level and the band Macroeconomic Forecasts benefit entitlements of existing em­ width, takes into account the possi­ and Microeconomic ployees would, however, require bility of a zero desired level, con­ Forecasters substituting explicit government strains the band to be nonnegative, Owen Lamont debt for the equally large implicit and allows asymmetric bands. NBER Working Paper No. 5284 debt of the unfunded program. The model is estimated on a October 1995 This paper shows that such a novel dataset that contains infor­ JEL Nos. D82, E17 debt-financed transition from an mation on both stock values and Monetary Economics unfunded program to a funded automobile expenditure for a large program is not just a change of number of households observed In the presence of principal­ form without economic effects. over a period of a year. The (5,s) agent problems, published macro­ Such a debt ..:financed transition rule is specified in terms of the ratio economic forecasts by professional would raise economic welfare if of car stock to nondurables. The economists may not measure three conditions are met: 1) the shortcuts usually employed in the expectations. Forecasters may use marginal product of capital ex­ empirical literature on (5,s) rules their forecasts to manipulate beliefs ceeds the rate of economic growth; can be avoided thanks to the rich­ about their ability. I test a cross­ 2) the capital intensity of the econ­ ness of the dataset and the rigorous sectional implication of models of omy is below the welfare-maximiz­ specification of the stochastic model. reputation and revelation of infor­ mation. I find that as forecasters ing level (that is, the marginal prod­ Having estimated the model and uct of capital exceeds the appro­ considered "goodness of fit" mea­ become older and more estab­ priate consumption discount rate); sures, I consider aggregation is­ lished, they produce more radical and 3) the rate of economic growth sues. First, I present a number of forecasts. Since these more radical is positive. negative results. Then, I consider forecasts in general are less accu­ illustrative calculations based on several simulations aimed at evalu­ rate, the accuracy of ex post fore­ the u.s. experience since 1960 sug­ ating the effects induced by inertial casts grows significantly worse as gest that the present value of the behavior on aggregate dynamics. forecasters become older and more gain from a debt-financed transi­ established. These findings indicate tion to a funded program would Is Regionalism Simply that reputational factors are at work exceed substantially the current a Diversion? Evidence in professional macroeconomic level of GDP. More explicitly, even from the Evolution forecasts. with a relatively high real con­ of the EC and EFTA sumption discount rate of 4.4 per­ Tamim Bayoumi and Models of Currency cent, the present value gain would Barry Eichengreen Crises with Self­ be about 1.5 dollars per dollar of NBER Working Paper No. 5283 Fulfilling Features current net social security wealth, October 1995 Maurice Obstfeld or about $17 trillion. JEL Nos. FO, Fl NBER Working Paper No. 5285 International Trade and Investment October 1995 JEL Nos. F33, E58 Consumer Durables and This paper considers the impact International Finance and Inertial Behavior: on trade of preferential arrange­ Macroeconomics, International Estimation and ments in Europe since the 1950s. (S,s) ~u1es Trade and Investment Aggregation of Using a first-difference version of Orazio P. Attanasio the gravity model, we find that the The discomfort a government NBER Wolking Paper No. 5282 EC and EFTA altered the pattern of suffers from speculation against its September 1995 international trade. We also find currency determines the strategic JEL Nos. C5, E2 evidence of trade diversion in sev­ incentives of speculators and the Economic Fluctuations eral cases, notably that of the EC in scope for multiple currency-market This paper presents an (5,s) the 1960s. equilibriums. After describing an il­ model for automobile consumption lustrative model in which high un­ and estimates it using a dataset of employment may cause an ex­ U.S. households. The model allows change rate crisis with self-fulfilling for unobserved heterogeneity in features, this paper reviews some

72. NBER Reporter Fall 1995 • other self-reinfor~ing ~echanisms. Individual saving through target­ widely in the literature also is re­ .. Recent econometnc eVidence seems ed retirement saving accounts­ jected. The only surviving evidence to support the practical importance IRAs and 401(k)s-grew rapidly in of any effect of schooling quality is of these mechanisms. the United States during the 1980s. in the return to college education. The microeconomic evidence pre­ We also test and reject convention­ Sectoral Solow Residuals sented in this paper suggests that al efficiency unit models of the Craig Burnside, most of the contributions to these pricing of labor services. The em­ Martin Eichenbaum, and programs represent new saving pirically concordant model of earn­ Sergio T. Rebelo that would not have occurred oth­ ings is a model of heterogeneous NBER Working Paper No. 5286 erwise. We compare the micro evi­ human capital in which regional October 1995 dence with macro saving, mea­ shocks affect the prices of less­ JEL Nos. D2, E3, 04 sured by National Income and skilled labor. Economic Fluctuations Product Accounts and Flow of Funds data. Stochastic Regime This paper presents measures of Switching and Stabilizing technology shocks corrected for The Schooling Quality­ Poli~ies Within Regimes capital utilization for aggregate and Karen K. Lewis disaggregated (two-digit Standard Earnings Relationship: Using Economic Theory NBER Working Paper No. 5289 Industrial Classification code) in­ to Interpret Functional October 1995 dustries. We correct for variations JEL Nos. F31, G12 in capital utilization by employing Forms Consistent Asset Pricing, International Finance industrial electrical use as a mea­ with the Evidence Heckman, and Macroeconomics sure of capital services. In contrast, James J. Anne Layne-Farrar, and This paper describes a class of the standard measures of technolo­ Petra Todd gy shocks used in the Real Busi­ stochastic stabilizing poliCies within NBER Working Paper No. 5288 ness Cycle (RBC) literature are asset price regimes that can be in­ October 1995 based on economywide data, and corporated easily into the frame­ JEL Nos. ]3~, 121, H40 assume that capital services are work of regime switching pro­ Labor Studies proportional to the stock of mea­ posed recently by Froot and Obst­ sured capital. To assess the impact This paper investigates the eco­ feld (1991). In contrast to previous of these differences, we contrast nomic and empirical foundations treatments of market-driven funda­ selected properties of the compet­ of the evidence relating earnings to mentals within the regime, authori­ ing technology shock measures. schooling quality. We replicate the ties stochastically counteract move­ We argue that the properties of Card-Krueger model for Census ments in the fundamentals before technology shocks for the manu­ years 1970, 1980, and 1990, and asset prices reach boundary points. facturing sector are quite different find that it consistently produces a This paper describes how the stabi­ from those used in the RBC litera­ strong relationship between school­ lizing intraregime intervention poli­ ture. We also find that correcting ing quality and the rate of return to cies can be used to characterize for capital utilization has important schooling. We then test key identi­ the behavior of monetary authori­ implications for the properties of fying assumptions used by Card ties before fixing an exchange rate, the Solow residual. and Krueger and others, and reject as studied by Flood and Garber several of them. When the assump­ (983). An intervention policy The Effects of Special tions are relaxed, the evidence for within target zone bands consistent Saving Programs on a strong effect of schooling quality with the empiric~l evidence also Saving and Wealth on earning is weakened greatly. belongs to this class of policies. Furthermore, the stylized features James M. Poterba, One crucial identifying assump­ Steven F. Venti, and tion is the absence of selective mi­ of these intervention policies may David A. Wise gration on the basis of earnings. be matched to actual data in a nat­ ural way. NBER Working Paper No. 5287 Nonparametric tests strongly reject October 1995 this hypothesis. The conventional ~ JEL Nos. J14, J28 assumption of linearity of the earn­ , Aging, Public Economics ings-schooling relationship used

NBER Reporter Fall 1995 73. Consumption Taxes: of individuals to conform to the We interpret the sign switch in Some Fundamental behavior of others in an environ­ their covariation with output as re­ Transition Issues ment of noncooperative decision­ flecting the interaction of our ad­ David F. Bradford making. Multiple equilibriums in justment cost specification with the NBER Working Paper No. 5290 these models, which are equivalent operation of two shocks: one that October 1995 to multiple self-consistent means affects the demand for equity, and JEL Nos. H25, H24, H20 for average choice behavior, will another that shifts the technology Public Economics exist when the social interactions for producing investment goods. exceed a particular threshold. We The adjustment costs cause the two A number of tax reform plans also study local stability of these prices to respond differently to under discussion in the United multiple equilibriums. We contrast these two shocks, and that is why States would replace the existing the properties of the noncoopera­ it is possible to choose the shock hybrid income-based system with a tive economy with the properties variances to reproduce the sign consumption-based system. In this of an economy in which a social switch. paper, I use uniform (that is, sin­ planner determines the set of indi­ These features of the model are gle-rate) consumption and income vidual choices. Additionally, we incorporated into a modified ver­ taxes to: 1) explain how the prob­ show the model to be well suited sion of a model analyzed in Bol­ lem of taxing "old savings" or "old to explaining a number of empiri­ drin, Christiano, and Fisher (1995). capital" manifests itself in the shift cal phenomena, such as threshold That model incorporates assump­ from an income to a consumption effects in individual behavior, ethnic tions designed to help account for base; 2) indicate the trade-offs that group fixed effects of income equa­ the observed mean return on risk­ must be confronted in dealing with tions, and large cross-group dif­ free and risky assets. We find that this phenomenon; 3) show how ferences in binary choice behavior. the various modifications not only price level changes that mayor account for the sign switch, but may not accompany a transition af­ also continue to account for the sa­ fect the distribution of gains and lbbin's q and Asset lient features of mean asset returns. losses; 4) sketch out how a transi­ Returns: Implications for tion might affect interest rates and Business Cycle Analysis We turn to the business cycle asset prices (including owner-occu­ Lawrence J. Christiano and implications of our model. The pied housing); 5) explore the case Jonas Fisher model does as well as standard in equity for protecting the tax-free NBER Working Paper No. 5292 models with respect to convention­ al business cycle measures of vola­ recovery of old savings; and 6) em­ October 1995 phasize the incentive problems that JEL Nos. E3, E37 tility and comovement with output, and on one dimension the model arise if savers and investors antici­ Economic Fluctuations significantly dominates standard pate a change in the tax rate in a The marginal cost of plant capa­ consumption-based system. models. The factors that help it ac­ city, measured by the price of count for prices and rates of return equity, is significantly procyclical. Discrete Choice with on assets also help it account for Yet the price of a major intermedi­ the fact that employment across a Social Interactions I: ate input into expanding plant ca­ Theory broad range of sectors moves to­ padty-investment goods--is coun­ gether over the cycle. Steven N. Durlauf and tercyclical. The ratio of these prices William A. Brock is Tobiil's q. FollOwing convention, The Child Care Industry: NBER Working Paper No. 5291 we interpret the fact that Tobin's q October 1995 Cost Functions, Efficiency, differs at all from unity as reflecting and Quality JEL Nos. D6, J2, R1 diminishing returns to expanding H. Naci Mocan Economic Fluctuations plant capadty by installing invest­ NBER Working Paper No. 5293 This paper analyzes aggregate ment goods ("adjustment costs"). October 1995 behavior when individual utility However, the phenomenon that in­ JEL Nos. D24, J13, J23 exhibits social interaction effects. terests us is not just that Tobin's q Health Economics We study generalized logistic mod­ differs from unity, but also that its els of individual choice that incor­ numerator and denominator have Using a newly compiled dataset, porate terms reflecting the desire such different cyclical properties. this paper provides insights into

74. NBER Reporter Fall 1995 • the characteristics of the child care Aging, Health Care Scale Economies, Returns " industry. First, there is no differ­ We use a new dataftle of insur­ to Variety, and the ence in the average quality of ser­ ance claims under employer-pro­ Productivity of Public vices produced by nonprofit versus vided health plans to describe the Infrastructure for-profit centers. Thus nonprofit pattern of expenditures in a large Douglas Holtz-Eakin and status is not a signal of higher corporation: Who spends? On Mary E. Lovely quality. what? For how long? The descrip­ NBER Walking Paper No. 5295 Second, the hypothesis of rela­ tion is illustrative of detail that can October 1995 tive inefficiency of nonprofit cen­ be distilled for other firms. JEL Nos. H40, H73, R13 ters with respect to for-profits is There are several noticeable fea­ Public Economics unfounded. On the other hand, tures of spending: 1) The differ­ We examine the productivity of centers that receive public money, ences in the provisions of the two public infrastructure in a general either from the state or federal gov­ FIRM plans seem to have substan­ equilibrium context. In our model, ernment, which is tied to higher tial effects on health care expendi­ infrastructure lowers costs in a standards, have variable costs that tures. Inpatient 'expenditures are a manufacturing sector characterized are 19 percent higher than other much larger fraction of the total by both firm-level returns to scale centers. under a first-dollar coverage plan and industry-level external returns Child care workers with 13 to 15 available to hourly employees. 2) to variety. Infrastructure alters fac­ years of education and workers Substance abuse and other mental tor prices, intermediate prices, and with 16 and more years of educa­ health disorders account for a sur­ the allocation of factors across sec­ tion are substitutes. Both of these prisingly large fraction of health tors. The effect on manufacturing groups are complements to work­ care costs. 3) The components of or aggregate output, however, is ers with 12 years of education or health care cost differ substantially indeterminate. In particular, our less. Centers have inelastic demand across individuals, demographic theory suggests that the degree of for workers. groups, and plans, suggesting that monopoly power influences public • There are economies of scale in further analysis of the "micro" de­ capital's productivity effect. " production. Controlling for the lev­ tails of medical expenditures may We test the model using state­ el of quality of services, a 10 per­ help to assess the probable impli­ level panel data. We confirm the cent increase in hours of children cations of alternative insurance re­ absence of direct effects on output, served brings about only an 8.5 forms. 4) There is substantial per­ but find suggestive evidence of a percent increase in costs in the sistence in individual health care positive impact of public capital on long run. There is no evidence of expenditures from year to year. Il­ manufacturing variety as measured economies of scope, though. Serv­ lustrative calculations suggest that by the number of manufacturing ing various age groups jointly is this persistence may lead to con­ establishments. These results indi­ not more efficient than serving siderable inequality in individual cate the need for future research them separately, although the issue costs under demand-side insurance on potentially important indirect is less clear in the case of preschool­ reforms such as those coupled with channels by which public capital ers and school-aged children. individual health account saving affects manufacturing productivity. The cost of increasing the quali­ plans, and thus may place impor­ ty of an average center from medi­ tant limitations on plans that other­ wise might be attractive. Measuring Poverty ocre to good is between 12 and 16 Among the Elderly cents per hour per child. The analysis in this paper sug­ gests that firm data can provide im­ Angus Deaton and portant insights into the black box Christina H. Paxson Where the Money Goes: NBER Working Paper No. 5296 Medical Expenditures in of expenditures under private in­ a Large Corporation surance plans. Hence these data October 1995 JEL Nos. J14, 132 Mark B. McClellan and may provide a greater understand­ Aging David A. Wise ing of how plan reforms may affect NBER Working Paper No. 5294 expenditures. Poverty counts are counts of in­ A October 1995 dividuals in poverty but are calcu­ • JEL Nos. J14, 110 lated from household or family

NBER RepotterFall 1995 75. data on income or expenditure. tion-fmanced system of delivery for tions are rolled over into IRAs or The transition from one to the oth­ mental health services developed other retirement saving plans, large er requires assumptions about in­ in Rochester, New York. We imple­ distributions are substantially more trahousehold allocation, about dif­ ment a cost-benefit analysis of the likely to be saved than smaller ferences in needs across different treatment program with three years ones are. Consequently, more than people, and about the extent of of data and program evaluation half of the dollars paid out as economies of scale. The number of techniques. We compare patient lump-sum distributions are rein­ elderly in poverty, or the number outcomes across randomly as­ vested. We also explore the corre­ of children in poverty, is sensitive signed study groups as well as lation between various individual to these assumptions and to differ­ across enrollment status. We find characteristics and the probability ences in living arrangements across that patients enrolled in the capita­ of rolling over a lump-sum distri­ age groups. We explore the sensi­ tion program do experience signifi­ bution. This is a first step toward de­ tivity of poverty counts to varia­ cantly lower costs without becom­ veloping a model that can be used tions in assumptions about child ing sicker, even after controlling to evaluate the long-term effects of costs and economies of scale using for attrition and sample selection. lump-sum distributions, or polides data from the United States and that might affect them, on the fi­ from six large states in India. Lump-Sum Distributions nancial status of elderly households. Because living arrangements of from Retirement Saving the elderly are so different in the Plans: Receipt The Underrepresentation United States and India, the use of and Utilization of Women in Economics: the latter forces us to think about James M. Poterba, A Study of Undergraduate household structure and poverty in Steven F. Venti, and Economics Students the United States. We argue that the David A. Wise Karen E. Dynan and offidal poverty counts in the Unit­ NBER Working Paper No. 5298 Cecilia Elena Rouse ed States are compromised by un­ October 1995 NBER Working Paper No. 5299 realistically high costs of children Aging, Public Economics October 1995 and by unrealistically high econo­ One of the central issues in JEL No. A2 mies of scale. We discuss how evaluating the ongoing shift from Labor Studies economies of scale and child costs defined-benefit to defined-contri­ Although women are underrep­ can be estimated from the data, us­ bution pension plans is the degree resented in the field of economics, ing identifying assumptions that la­ to which assets in defined-contri­ many people see little need for in~ bel private goods and adult goods, bution plans will be withdrawn be­ tervention, arguing that women are and we make calculations based fore plan partidpants reach retire­ inherently less interested in eco­ on the 1990 Consumer Expenditure ment age. The annual flow of with­ nomics, or are less willing or able Survey. We obtain plausible esti­ drawals from such plans, which are to acquire the math skills needed mates of child costs, together with known as lump-sum distributions to do well in the subject. At the a number of interesting but hard­ and which are frequently but not same time, others support active to-explain anomalies, when we try always associated with employ­ efforts to increase the number of to estimate economies of scale. ment changes, has exceeded $100 women in the field, pointing to billion in recent years. This flow is other possible causes of their cur­ Evaluating Mental Health substantially greater than the flow rent underrepresentation. These Capitation Treatment: of new contributions to IRAs and people argue, for example, that Lessons from Panel Data other targeted retirement saving women are deterred from entering Debra Sabatini Dwyer, programs. the field because of a lack of fe­ Olivia S. Mitchell, This paper draws on data from male role models, or that women Robert Cole, and the 1993 Current Population Survey are discouraged by an unappealing Sylvia K. Reed and the Health and Retirement Sur­ classroom environment. This study NBER Working Paper No. 5297 vey to summarize the incidence attempts to evaluate these hypothe­ October 1995 and disposition of lump-sum distri­ ses. We examine the factors that Health Care butions. We find that while less influence undergraduate students' This paper evaluates a capita- than half of all lump-sum distribu- decisions to become economics

76. NBER Reporter Fall 1995 • majors by analyzing a survey of the fiscal authority is impatient, a is simply too little life-cycle saving . students in the introductory eco­ money-based stabilization provides for higher growth to have a large nomics course at Harvard Universi­ more fiscal discipline and higher effect on the aggregate saving rate. ty, as well as data on an entire class welfare for the representative agent The habit formation model also im­ of students from Harvard's registrar. than an exchange rate-based stabi­ plies very small effects of growth on We fmd that although women in lization does. Data for Latin Ameri­ saving rates. A large portion of the the introductory economics course ca stabilizations in the last quarter­ observed cross-country correlation at Harvard tend to begin the century seem to' confirm the notion between saving and growth cannot course with a weaker math back­ that stabilizing by using money be explained by these models. ground than men, this does not ap­ rather than the exchange rate helps pear to explain much of the gender induce politicians to reduce the fis­ Labor Market Effects of difference in students' decisions cal deficit. Spinal Cord Injuries about whether to major in eco­ in the Dawn of nomics. The class environment and Saving and Growth: the Computer Age the presence or absence of role Evidence from Microdata Alan B. Krueger, models also do not explain much Christina H. Paxson Douglas Kruse, and of the gender gap. On the other NBER Working Paper No. 5301 Susan Drastal hand, women do less well in eco­ October 1995 NBER Working Paper No. 5302 nomics relative to their other JEL No. E21 October 1995 courses than men do, and control­ Aging JEL No. J28 ling for this difference in relative Labor Studies performance significantly diminish­ This paper examines whether What effect does a severe dis­ es the estimated gender gap. An the observed cross-country correla­ ability have on individuals' em­ economically large, but statistically tion between aggregate saving ployment and earnings? Has the insignificant, difference between rates and economic growth can be computer revolution lessened the sexes in the probability of majoring explained by models in which adverse labor market conse­ in economics remains, however. higher growth increases saving quences of severe disabilities? This This remaining gender gap may be rates, rather than the other way paper investigates the labor market •caused by differing tastes or infor­ around. The paper focuses on two effects of severe, traumatic disabili­ mation about the nature of eco­ explanations of why growth might increase saving. First, standard life­ ties resulting from spinal cord inju­ nomics.· As evidence, we find that ries (SCIs). We compare the em­ women who were considering ma­ cycle theory implies that higher ployment experiences of a sample joring in economics when they be­ growth will increase the lifetime of individuals with SCIs to those of gan introductory economics were wealth of younger savers relative former coworkers over the same about as likely to choose econom- to older dissavers, thereby increas­ period, and to two random sam­ ics as men were. ing the aggregate saving rate. Sec­ ond, models of consumption with ples of individuals in New Jersey. Money-Based Versus habit formation imply that con­ The analysis is based in large part Rxchange Rate-Based sumption responds slowly to unex­ on a 1994 telephone survey of Stabilization with pected income growth; thus unan­ New Jersey adults who had SCls Endogenous Fiscal Policy ticipated growth can produce a within the past ten years. Results higher saving rate, at least in the indicate that the occurrence of an Aaron Tornell and short run. I assess the validity of SCI causes a steep decline in em­ Andres Velasco these explanations using time se­ ployment, hours worked, and NBER Working Paper No. 5300 ries of cross-sections of household weekly earnings, but relatively little October 1995 income and consumption surveys change in wage rates for those International Finance and from four countries: the United who work. The computer revolu­ Macroeconomics States, Britain, Taiwan, and Thai­ tion has the potential to expand We present a standard intertem­ land. I find that in three out of the employment opportunities for peo­ poral model in which fiscal policy four countries there is evidence ple with disabilities. Our results in­ is determined by an optimizing but that saving behavior is consistent dicate that having computer skills •nonbenevolent fiscal authority. If with the life-cycle theory, but there is associated with higher earnings, NBER Reporter Fall 1995 77. a faster return to work, and faster on parameters such as endow­ that complementarities are impor­ earnings recovery for SCI individu­ ments and technology. tant to aggregate volatility, even af­ als, after holding constant other va­ ter I remove observable aggregate riables such as education. There is Comovement in Cities shocks from the data. Local spill­ no apparent earnings gap between John Shea overs are -particularly important, SCI and non-SCI computer users; NBER Working Paper No. 5304 explaining between 15 and 36 per­ among those who do not use com­ October 1995 cent of the volatility of manufactur­ puters at work, however, the earn­ JEL Nos. E32, Rll ing employment. ings of SCI employees lag behind Monetary Economics those of non-SCI employees. De­ Generic Entry and the spite the benefits, individuals with Recent research has shown that Pricing of Pharmaceuticals industries that locate together in SCls are less likely to use comput­ Richard G. Frank and ers than the general population, space also move together over the David S. Salkever business cycle, and that this corre­ though. NBER Working Paper No. 5306 spondence between spatial and October 1995 temporal comovement is important The Antebellum. Health Care Transportation to aggregate volatility. This paper Revolution and asks whether the correspondence During the 1980s the share of Factor-Price Convergence is attributable to local common prescriptions for generic products Matthew J. Slaughter shocks or to local spillovers. I ex­ sold by retail pharmacies roughly doubled. The price response of NBER Working Paper No. 5303 amine interindustry co movements brand-name products to generic October 1995 within seven large U.S. cities, and entry has been a source of contro­ JEL Nos. F15, J31, N31 fmd strong evidence of local spill­ versy, though. In this paper we es­ International Trade and Investment overs. I estimate that local spill­ overs explain roughly one-third of timate models of price responses to In antebellum America an exten­ manufacturing employment volatili­ generic entry in the market for sive network of canals and rail­ ty at the city level. Local spillovers brand-name and generic drugs. We roads was constructed that slashed do not appear to result from trans­ study a sample of 32 drugs that lost transportation costs across regions. port costs and locally tr-aded goods. patent protection during the early This "transportation revolution" to mid-1980s. Our results provide presents an interesting case study Complementarities strong evidence that brand-name of the factor-price convergence and Comovements prices increase after entry, and are (FPC) theorem. In this paper I John Shea accompanied by large decreases in study the extent to which com­ NBER Working Paper No. 5305 the price of generic drugs. modity prices and factor prices October 1995 converged across regions between JEL Nos. E32, R12 Emerging Equity 1820 and 1860. I find very little evi­ Monetary Economics Market Volatility dence of antebellum FPC across re­ Geert Bekaert and gions. I further fmd that commodi­ Short-run interindustry comove­ Campbell R. Harvey ty prices equalized quite markedly, ment may be attributable either to NBER Working Paper No. 5307 but that nominal labor prices common shocks or to complemen­ October 1995 equalized very little, if at all Given tarities· that propagate shocks JEL Nos. F3, GO, G 1, C5 this result, I go on to discuss two across sectors. This paper assesses Asset Pricing aspects of the antebellum economy the importance of input-output that probably helped prevent FPC: linkages, and spillovers from ag­ Returns in emerging capital mar­ differences across regions in en­ gregate and local activity to co­ kets are very different from returns dowments and technology. This movement in postwar U.S. manu­ in developed markets, While most finding underscores the fact that facturing. I find that input-output previous research has focused on the FPC theorem does not have linkages and local activity spill­ average returns, we analyze the unambiguous empirical predic­ overs are important to comove­ volatility of the returns in emerging tions. How commodity prices feed ment, while aggregate activity spill­ equity markets. We characterize into factor prices depends crucially overs are not important. I also find the time-series of volatility in

78. NBER Reporter Fall 1995 '.emerging markets, and explore the ments in human development indi­ domestic special interest groups. A ". distributional foundations of the cators, but it does increase the size framework of this sort is essential variance process. Of partiCular in­ of government. I also find that the for a proper analysis of a host of terest are asymmetries in volatility impact of aid does not vary accord­ important problems, such as nego­ and the evolution of the variance ing to whether recipient govern­ tiations about tariff levels or the process after periods' of capital ments are liberal democratic or formation of free trade areas. Re­ market reform. We shed indirect highly repressive. But liberal politi­ cent studies have developed suit­ light on the question of capital cal regimes and democracies, cet­ able tools for this purpose. market integration by exploring the eris paribus, have on average 30 changing influence of world factors percent lower infant mortality than International Cycles on the volatility in emerging mar­ the least free regimes. This may be Stephen G. Cecchetti and kets. Finally, we investigate the because of greater empowerment Ani! K Kashyap cross-section of volatility. We use of the poor under liberal regimes, NBER Working Paper No. 5310 measures such as asset concentra­ even though the political elite con­ October 1995 tion, market capitalization to GDP, tinues to receive" the benefits of aid JEL Nos. E23, E32 size of the trade sector, cross-sec­ programs. One implication is that Monetary Economics tional volatility of individual securi­ short-term aid t;rgeted to support We study 20 years of monthly ties within each country, turnover, new liberal regimes may be a more production data for 11 manufactur­ foreign exchange variability, and successful means of reducing pov­ ing industries in 19 countries. Us­ national credit ratings to character­ erty than current programs. ing the fact that in some countries ize why volatility is different across production virtually shuts down for emerging markets. Politics and Trade Policy one summer month, together with Elhanan Helpman the differences in the timing of ag­ Politics and the NBER Working Paper No. 5309 gregate cyclical fluctuations, we are " Effectiveness of October 1995 able to learn about the cost struc­ . Foreign Aid JEL Nos. F13, F15 • ture of different industries. Our pri­ Peter Boone International Trade and Investment, mary finding is that during a boom NBER Working Paper No. 5308 International Finance and year, summer shutdowns are short­ October 1995 Macroeconomics er. Rather than increasing produc­ JEL Nos. El, F35, 011 First I describe a number of po­ tion further during the rest of the International Finance and litical economy approaches that year, producers reallocate activity Macroeconomics have been developed to explain from high-output months to low­ Critics of foreign aid programs trade policies. I present all the ap­ output months. long have argued that poverty re­ proaches in a unified framework We also find that there are im­ flects government failure. In this that helps to show off the key dif­ portant seasonaVcyclical interac­ paper I analyze the effectiveness of ferences among them. These com­ tions common to all industries foreign aid programs to gain in­ parisons revolve around tariff for­ within a given country, and that sights into political regimes in aid­ mulas that are predicted by politi­ these countries' effects are larger recipient countries. My analytical cal equilibriums. A typical formula than the pure industry effects. The framework shows how three styl­ explains cross-sectoral variations in correlation of the cross-country dif­ ized politicaVeconomic regimes, la­ rates of protection as well as differ­ ferences with measures of taxation beled egalitarian, elitist, and lais­ ences in average rates of protection and labor market structure suggests sez-faire, would use foreign aid. I across countries. Then I review a the possibility th.at differences in then test reduced-form equations set of results that emerge from a the willingness (and ability) to sub­ using data on nonmilitary aid flows new approach to the interaction of stitute labor intertemporally are re­ to 96 countries. I find that models international economic relations sponsible for the variation. of elitist political regimes best pre­ with domestic politics. Importantly, dict the impact of foreign aid. Aid there are two-way interactions in does not significantly increase in­ such systems. They link the forma­ vestment and growth, nor benefit tion of trade policies in the interna­ I the poor, as measured by improve- tional arena with the activities of

NBER Reporter Fall 1995 79. Good News for Value error. I then discuss the relation­ Validating the conjecturalG Stocks: Further Evidence ship of uncovered interest parity to Variation Method: The on Market Efficiency real interest parity, and the implica­ Sugar Industry, 1890-1914 Rafael Ia Porta, tions of uncovered interest parity David Genesove and Josef Lakonishok, for cointegration of various quanti­ Wallace P. Mullin Au1dreiShleHe~and ties. I also survey the modeling NBER Working Paper No. 5314 Robert W. VlShny and testing for risk premiums. In­ October 1995 NBER Working Paper No. 5311 cluded in this area are tests of the JEL No. L13 October 1995 consumption CAPM, the latent vari­ Industrial Organization Assset Pricing able model, and portfolio-balance models of risk premiums. Further, I The Conjectural Variation (CV) This paper examines the hy­ methodology uses the responsive­ examine general equilibrium mod­ pothesis that the superior return to els of the risk premium and ex­ ness of price to cost determinants under differing demand conditions so-called value stocks is the result plore their empirical implications. of expectational errors made by in­ to infer market power and cost. The survey does not cover the im­ vestors. We study stock price reac­ portant areas of learning and peso Thus it substitutes demand infor­ mation for complete cost informa­ tions around earnings announce­ problems, tests of rational expecta­ tion. We use the American sugar ments for value and glamour tions based on survey data, or the refining industry at the turn of the stocks over a five-year period after models of irrational expectations century to assess the efficacy of the portfolio formation. The announce­ and speculative bubbles. ment returns suggest that a signifi­ CV approach. To do so we com­ cant portion of the difference in re­ pare direct measures of marginal turn between value and glamour Liquidity Models in Open cost and price-cost markups with Economies: Theory and the indirect estimates obtained stocks is attributable to earnings Empirical Evidence surprises that are systematically from the CV method. We find that Nouriel Roubini and more positive for value stocks. The the CV method performs reason­ Vittorio U. Grilli evidence is not consistent with a ably well. It yields estimates of in­ NBER Working Paper No. 5313 risk-based explanation for the re­ dustry conduct that are close to the October 1995 turn differential. direct measure we derive from full JEL Nos. E52, F41 cost information, and robust to the International Finance and choice of the functional form of The Forward Discount Macroeconomics Anomaly and the Risk demand. The conduct parameter is Premium: A Survey of This paper presents an overview underestimated, but the deviation Recent Evidence of recent theoretical and· empirical is minimal in our context. The CV Charles M. Engel research on "liquidity models" in methodology. does a better job of detecting differences in conduct NBER Working Paper No. 5312 open economies: a class of opti­ mizing models in which money has arising from different structural re­ October 1995 effects on real asset prices and eco­ gimes (corresponding to the after­ JEL No. F3 nomic activity without the "ad-hoc" math of entry), but only when the International Finance and assumption of price/wage sticki­ researcher imposes the (a priori Macroeconomics, Asset Pricing ness. The nonneutrality of money known) restriction of cost stability. The unbiasedness of the for­ derives from a temporary segmen­ ward exchange rate is rejected in tation between goods and asset Unemployment Insurance tests from the current floating ex­ markets. After surveying the theo­ and Household Welfare: change rate era. This paper surveys retical literature on liquidity mod­ Microeconomic Evidence advances in this area since the els, we present empirical evidence 1980-93 publication of Rodrick's (987) sur­ based on vector autoregression Daniel S. Hamermesh and vey. It documents that the change econometric techniques for the Daniel T. Slesnick in the future exchange rate gener­ seven major industrial countries. NBER Working Paper No. 5315 ally is related negatively to the for­ This evidence is consistent with the October 1995 ward discount. I review properties main implications of the liquidity JEL Nos. 138, D12 of the expected forward forecast models. Labor Studies

80. NBER RepotterFall 1995 • This study examines the relative functions as a large tax on savings. Collapsing Exchange Rate economic well-being-as measured College financial aid also functions Regimes: Another by consumption flows that are d~­ as an income tax. This paper esti­ Linear Example rived from information on house­ mates the size and determinants of Robert P. Flood, holds' spending in the Consumer these income and asset taxes. We Peter M. Garber, and Expenditure Surveys from 1980-93 find that the marginal income tax Charles Kramer -of households that receive un­ typically ranges from 2 percent to NBER Wolking Paper No. 5318 employment insurance CUD bene­ 16 percent and the marginal asset October 1995 fits. For each quarter during this levy from somewhat under 10 per­ JEL Nos. F30, F33 period we obtain the per-capita cent to as high as 25 percent. If a International Finance and and equivalence-scale adjusted typical family chooses to accumu­ Macroeconomics economic welfare of the two types late $100,000 in assets rather than of households. Adjusting for differ­ consuming these resources, it loses In the literature on speculative ences in the households' character­ fmancial aid worth $10,000-$20,000. attacks on a fixed exchange rate, it istics, we find: 1) the average UI usually is assumed that the mone­ recipient household during this pe­ tary authority responsible for fixing riod had a level of economic well­ the exchange rate reacts passively being that was on average between Growth Effects of Income to. the monetary disruption caused 3 and 8 percent below that of oth­ and Consumption Taxes: by the attack. This assumption is erwise identical households (de­ Positive and grossly at odds with actual experi­ pending on the welfare measure Normative Analysis ence in which monetary-base im­ used); 2) during a substantial part Nourie! Roubini and plications of the attacks usually are of this time the economic well-be­ Gian Maria Milesi-Ferretti sterilized. Such sterilization renders ing of households that received UI NBER Working Paper No. 5317 the standard monetary-approach benefits was at least that of other October 1995 attack model unable to provide in­ . households; and 3) there is no cyc­ JEL Nos. E6Z, 041 tellectual guidance for recent attack • lical variation in the relative well­ International Finance and episodes. In this paper we describe being of UI recipient households Macroeconomics, Public Economics the problems with the standard compared to others. These findings model and develop a version of We examine the effects of in­ imply that during the 1980s and the portfolio-balance exchange rate come and consumption taxation in early 1990s, states' UI programs did model that allows us to study epi­ the context of models in which sodes with sterilization. Sterilized a satisfactory job of maintaining the growth is driven by the accumula­ well-being of UI recipients. Emer­ attacks may be regarded as a labo­ tion of human and physical capital. ratory test of the monetary versus gency programs enacted during re­ We discuss the different channels the portfolo-balance model of ex­ cessions raised potential duration through which these taxes affect sufficiently to prevent the econom­ change rates. The monetary model economic growth, and show that in fails the test. These issues are moti­ ic position of the average UI recipi­ general the taxation of factor in­ vated by reference to the December ent from deteriorating. comes (human and physical capi­ 1994 collapse of the Mexican peso. tal) reduces growth. The effects of The Implicit Taxes from consumption taxation on growth College Financial Aid depend crucially on the elasticity The Impact of Price, Andrew W. Dick and of labor supply, and therefore on Availability, and Alcohol Aaron S. Edlin the specification of the leisure ac­ Control Policies on Binge NBER Wolking Paper No. 5316 tivity. We also derive implications for Drinking in College October 1995 the optimal intertemporal choice of FrankJ. Chaloupka and Public Economics tax instruments. Henry Wechsler NBER Wolking Paper No. 5319 Families who heed the advice of October 1995 "experts" and save for their chil­ dren's college education typically JEL No. 11 Health Economics ... receive less finandal aid. The vari­ ation in the net price of college • We estimate the effects of beer

NBER ReponerFall 1995 81. prices, the availability of alcohol, tive research efforts with sdentists In this paper, we derive three and polides related to driving un­ at other organizations, espedally at lessons from Mexico's experience. der the influence (of alcohol) on universities. Formal market con­ First, deep reforms such as trade drinking and binge drinking among tracts rarely are used to govern liberalization are not likely to hap­ youths and young adults. We use these exchanges of scientific knowl­ pen by government decree. In­ data from a nationally representa­ edge. Our findings suggest that the stead, they usually come about tive survey of students in U.S. col­ use of boundary-spanning social when the unanimous blocking of leges and universities. Our es­ networks by the two NBFs increas­ reform by powerful elites breaks timates indicate that the drinking es both their learning and their down. In the case of Mexico, this practices of college students are flexibility in ways that would not happened during a fiscal crisis, sensitive to the price of beer, with be possible within a self-contained when some groups tried to dis­ an average estimated price elastici­ hierarchical organization. place other groups in order to cap­ ty of drinking of -0.066 and an av­ ture a greater share of fiscal rev­ erage estimated price elasticity of Research and Productivity enue. Second, in the presence of binge drinking of -0.145. However, Boyan Jovanovic and entrenched elites, the sustainability when dividing the sample by gen­ Yaw Nyarko of reform depends on the exis­ der, we find that the effects of pri­ NBER Working Paper No. 5321 tence of new groups that benefit ces on drinking are limited to October 1995 from the new status quo and have young women. In addition, we find JEL Nos. 031, 033 enough power to defend it Thus, a significant negative relationship Productivity the speed of successful reform is determined by the speed with between the strength of policies re­ We model research as a Signal which new groups are consolidat­ lated to drinking and driving among on an unknown parameter of a youths and young adults and technology. We distinguish applied ed. Initially, Mexico limited radical drinking by college students. How­ from basic research, and show that liberalization to the manufacturing sector. The government only re­ ever, our results indicate that many finns in the same industry optimal­ cently has begun to undertake seri­ elements of campus life (including ly can choose different research ous liberalization in the services partidpation in a fraternity or soror­ portfolios. Basic research can seem and agriculture sectors. The third ity, living on campus, and the ready to have a higher rate of return than lesson we take from Mexico is that availability of alcoholic beverages) applied research, even though it the importance of formal agree­ are among the most important really doesn't: essentially, finns on ments such as NAFTA lies not so determinants of drinking and binge a "fast-track" upgrading policy opt much in the ability of these agree­ drinking among college students. for basic research, but fast- and slow-track upgrading policies can ments to reduce average import coexist in a long-run equilibrium. tariffs among their parties and im­ Social Networks, Learning, We also derive the lag structure prove their terms of trade vis a vis and Flexibility: Sourcing the rest of the world, as claimed by Scientific Knowledge in for how Rand D affects the firm's stock of knowledge. To a first ap­ the optimal tariff literature, but in New Biotechnology Firms proximation, the lags decay geomet­ that they serve as commitment de­ Julia Porter liebeskind, rically (as typically is assumed in vices to force reforms to continue. Amalya Lumerman Oliver, practice) but the rate of decay is en­ Lynne G. Zucker, and dogenous, and depends on how fast MarilylUl B. Brewer Exchange Rate Variability the firm is upgrading its technology. and the Riskiness of U.S. NBER Working Paper No. 5320; Multinational Firms: October 1995 The Political Economy of Evidence from the JEL No. 031 Mexico's Entry to NAFTA Breakdown Productivity Aaron Tornell and Eli Bartov, Gordon M. Bodnar, We examine how two highly Gerardo Esquivel and Aditya Kaut successful new biotechnology finns NBER Working Paper No. 5322 NBER Working Paper No. 5323 (NBFs) source their most critical in­ October 1995 October 1995 put: sdentific knowledge. We find International Trade and Investment, JEL Nos. F23, F31 that scientists at the two NBFs en­ International Finance and Asset Pricing, International Finance ter into large numbers of collabora- Macroeconomics and Macroeconomics

82. NBER RePorter Fall 1995 This study assesses the impact U.S.-Taiwan experience seems to Krugman framework, we fmd that: of exchange rate variability on the suggest a grossly negative answer. 1) the development of the EAEC by riskiness of u.s. multinational firms Bilateral negotiations for market the leadership of Malaysia would • by examining the relationship be­ opening with the threat of unilater­ be a natural response of Asian tween exchange rate variability and al trade sanctions (such as Section countries against two big blocs in stock return volatility, and by de­ 301 action) tend to encourage the world, EU and NAFTAj 2) it is composing this relationship into trade preferences, and U.S. nego­ natural for the United States to dis­ components of systematic and di­ tiators are inclined to accept such courage this move because the for­ versifiable risk. Focusing on two preferential arrangements in areas mation of an economic bloc in Asia periods around the 1973 switch where U.S. domestic interests are will have a negative economic im­ from fixed to floating exchange homogeneous and concentrated. pact on the non-Asian countriesj 3) rates, we find a significant increase Even in the case of tariff negotia­ it is natural for the United States to in the volatility of the monthly tions in which any tariff conces­ propose an opposing coalition stock returns of u.s. multinationals sions made by Taiwan are extend­ such as the APEC to nUllify the corresponding to the period of in­ ed to other trading partners on a possible economic impact of the creased exchange rate variability. most favored nation basis, bilateral­ EAECj but 4) perhaps the APEC This increase in stock return vola­ ism does not ~ecessarily enhance will be a good roundabout way to­ tility is also significant relative to multilateral principles. The scope ward international free trade. the increase in stock return volatili­ of tariff concessions made by Tai­ ty for firms in three control sam­ wan shows a strong bias in favor Inflation and ples. Using a single-factor market of the sectors in which the United Economic Growth model, we show that this increase States has a comparative advantage RobertJ. Barco in total volatility led to a significant in Taiwan's market and the sectors NBER Working Paper No. 5326 increase in market risk (beta) for the in which U.S. domestic industries October 1995 multinational firms relative to the exhibit monopoly power. Mean­ JEL Nos. 040, E31 control samples between the two while, U.S. commitments to GATT Economic Fluctuations, Growth, • ,. periods. Collectively, these results strengthen its position in bilateral Monetary Economics suggest that investors perceived the negotiations and help persuade increase in exchange rate variabili­ Taiwan, which is not a member of I use data for around 100 coun­ ty after 1973 to be associated with GATT, to make similar concessions. tries from 1960 to 1990 to assess an increase in the riskiness of cash the effects of inflation on economic flows of multinational firms that EU, NAFTA, and Asian performance. If a number of coun­ reguired compensation in terms of Responses: A Perspective try characteristics are held constant, higher expected returns. from the Calculus then the regression results indicate of Participation that the effects of an increase in Bilateral Negotiations Junichi 6oto and average inflation of 10 percentage and Multilateral Trade: Koichi Hamada points per year are a reduction in the growth rate of real per capita The Case of Taiwan-U.S. NBER Working Paper No. 5325 GDP by 0.2-0.3 percentage points Trade Talks October 1995 per year and a decrease in the ratio Tain-Jy Chen and JEL Nos. F12, F13, F15 of investment-to-GDP of 0.4-0.6 Meng~chun Uu International Trade and Investmant percentage points. Since the statis­ NBER Wolking Paper No. 5324 How should the Asian countries tical procedures use plausible in­ October 1995 cope with the formation of the EU struments for inflation, there is JEL No. F13 and NAFTA? Is it desirable for them some reason to believe that these International Trade and Investment to form their own trading area? relationships reflect causal influ­ This paper reviews the history And, if so, is it better to have a ences from inflation to growth and of bilateral trade negotiations be­ closed one, such as the EAEC, or a ihvestment. However, statistically tween Taiwan and the United more open one, such as the APEC? significant results emerge only States. The question posed at the Relying on public economics when high-inflation experiences outset is: does bilateralism enhance and the calculus of participation, are included in the sample. Al­

'> .' or jeopardize multilateralism? The combined with the Dixit-Stiglitz- though the adverse influence of in- •

NBER Reporter Fall 1995 83. flation on growth looks small, the model that yields a form of tax In standard models of the bal­ long-term effects on standards of smoothing as a basis for debt man­ ance of payments, crises occur living are substantial. For example, agement. The main analysis uses a when investors begin to doubt the a shift in monetary policy that rais­ tractable form of the one-sector credibiiity of the government's es the long-term average inflation stochastic growth model. The type commitment to its exchange rate rate by 10 percentage points per of taxation that yields the clearest policy. In this paper, we develop year is estimated to lower the level results on tax smoothing is a pro­ an alternative model in which bal­ of real GDP after 30 years by 4-7 portional levy on consumption. In ance-of-payments crises occur even percent, more than enough to justi­ a simple benchmark case, optimal if the credibility of government fis­ fy a strong interest in price stability. debt management entails the issue cal, monetary, and exchange rate of indexed consols. More general­ policies is never· in doubt. In this Optimal Debt Management ly, payouts on debt also would be alternative model, international contingent on aggregate consump­ RobertJ. Barro lending is constrained by the risk tion and the level of government of repudiation. Balance-of-pay­ NBER Working Paper No. 5327 spending. October 1995 ments crises occur when the gov­ JEL Nos. E62, H30 ernment and citizens of a country The Political Economy of hit their international borrowing Economic Fluctuations, Public State-Provided Health Economics, Growth constraints. Our model is broadly Insurance in the consistent with events in Mexico Optimal debt management can Progressive Era: Evidence from 1987-95. More generally, our be thought of in three stages. First, from California model suggests that countries that if taxes are lump sum and the oth­ DoraL Costa undertake sweeping macroeco­ er conditions for Ricardian equiva­ NBER Working Paper No. 5328 nomic and structural reforms lence hold, then the division of October 1995 should expect to face a balance-of­ government financing between JEL Nos. Ill, 118, N12 payments crisis when they exhaust debt and taxes is irrelevant, and DeVelopment of the American Economy their access to international capital the whole level of public debt is I investigate why the United inflows. indeterminate from an optimal-tax States did not adopt European-style standpoint. Second, if taxes are dis­ health insurance in the 1910s by Privatization of Social torting, then the timing of taxes examining voting determinants on Security: How It Works generally will matter; for example, the 1918 referendum on state-pro­ and Why It Matters it may be desirable to smooth tax vided health insurance in Califor­ Laurence J. Kotlikoff rates over time. This consideration nia. I find that although the persua­ NBER Working Paper No. 5330 makes determinate the levels of siveness of interest groups, espe~ October 1995 debt at various dates, but does not cially doctors, was an important JELNo. HO pin down the composition of the determinant of the 1918 vote, inter­ Public Economics debt, say. by maturity. Finally, if est groups alone could not explain there is uncertainty about real in­ the resounding defeat of state~pro­ This paper uses the Auerbach­ terest rates, levels of public outlay, vided health insurance. Voters Kotlikoff dynamic life-cycle model GDP, and so on, then the relation­ were unwilling to pass a costly (AK model) to examine the macro­ ship of tax rates to states of nature measure with an unpredictable economic and efficiency effects of becomes important. In some cases, outcome, I find. privatizing Social Security. It also optimal taxation dictates the uses a simple privatization propos­ smoothing of tax rates over states How Mexico Lost Its al, the Personal Security System, as of nature, and this element may Foreign Exchange Reserves a framework for discussing a num­ pin down the composition of the Andrew Atkeson and ber of other issues associated with debt. For example, the maturity Jose-Victor Rios-Rull the privatization of Social Security, structure can be designed to insulate NBER Working Paper No. 5329 including transition rules and the government's financing costs October 1995 changes in the overall degree of from shifts in real interest rates. JEL Nos. F32, F34, F41 progressivity. This paper studies dynamic opti­ International Finance and According to the AK model, pri­ mal taxation in an equilibrium Macroeconomics vatizing Social Security can gener-

84. NBER Reporter Fall 1995 • ate very major long-run increases Security's privatization could pro­ interpretations of the results of in output and living standards.· vide more survivors' protection studies using aggregate data to as­ These gains come largely, but not than the current system, as well as sess school input effects exclusively, at the expense of exist­ eliminating much of the current ing generations. Indeed, the pure system's seemingly capridous re­ Non-Keynesian Effects of efficiency gains from privatization distribution between two-earner Fiscal Policy Changes: can be substantial. Effidency gains and single-earner couples. But the International Evidence and here refers to the welfare improve­ proposal's analysis also suggests the Swedish Experience ment available to future genera­ that these benefits from privatiza­ Francesco Giavazzi and tions after existing generations tion must be set against a possible Marco Pagano have been compensated fully for reduction in progressivity and a NBER Worldng Paper No. 5332 their losses from privatization. The likely reduction in the amount of November 1995 precise size of the efficiency gain longevity insurance available to the International Finance and depends on the existing tax struc­ elderly through annuities. Macroeconomics ture, the linkage between benefits In earlier work we documented and taxes under the existing Social The Effect of Measured two episodes in which a sharp fis­ Security system, and the choice of School Inputs on cal consolidation was associated the tax instrument used to fmance Academic Achievement: with a surprisingly large expansion benefits during the transition. Evidence from the 1920s, in private domestic demand. In this When the initial tax structure 1930s, and 1940s paper we draw on further evidence features a progressivekincome tax, Birth Cohorts to investigate if and when changes when the existing system's benefit­ Susanna Loeb and John Bound in fiscal policy can have such non­ tax linkage is low, when consump­ NBER Working Paper No. 5331 Keynesian effects. In the first part tion taxation is used to finance So- November 1995 of the paper, we analyze cross­ cial Security benefits during the JEL Nos. 120, J31 country data for 19 OECD coun­ transition, and when existing gen­ Labor Studies tries. In the second part, we con­ • . . erations are compensated fully for This study uses data from the centrate on the Swedish fiscal ex­ their privatization losses, there is a National Opinion Research Center's pansion of the early 1990s. 4.5 percent simulated welfare gain General Social Surveys to explore The cross-country evidence on to future generations from privati­ the effects of measurable school private consumption confirms that zation. But if these drcumstances characteristics on student achieve­ changes in fiscal policy-both con­ don't hold, the efficiency gains ment. It differs from many other tractions and expansions-can from privatization are likely to be studies by its use of aggregate data have non-Keynesian effects if they smaller, possibly even negative. on older cohorts, usually associated are sufficiently large and persistent. For example, when the initial tax with research on the influence of It also suggests that these effects structure is a proportional income school inputs on earnings. Studies can result not only from changes in tax, when benefit-tax linkage is on earnings have tended to find public consumption, but also to perceived to be dollar for dollar, substantial effects, while much of some extent from changes in taxes when the income tax rate is raised to the research on achievement using and transfers. This latter result is finance Social Security benefits dur­ contemporary, cross-sectional data also consistent with the Swedish ing the privatization transition, and has not However, we find substan­ experience, in which a decrease in when current generations are fully tively large effects, similar in size net taxes (with almost no change compensated, there is a 3.1 percent to those found in many earnings­ in public consumption) was associ­ welfare loss to future generations. focused studies. In this way, our ated with a dramatic fall in private The illustrative Personal Security results point to the importance of domestic demand. System shows that there are simple aggregation and cohort effects in Our evidence and that of other ways to privatize the retirement modeling the relationship between studies concur that during the Swe­ portion of the U.S. Sodal Security school inputs and student out­ dish fiscal expansion of the early system and to credit workers for comes. In particular, the level of 1990s, a large negative error ap­ .~eir past Sodal Security contribu­ data aggregation appears impor­ pears in the consumption function. _ions. It also suggests that Social tant, bringing into question causal There is less consensus about how

NBER Repon-erFall 1995 85. this error should be interpreted, work teams, flexible job assign­ provide no option value to an ini­ but it is clear that the most obvious ments, employment security, train­ tial hirer. candidates (wealth effects and af­ ing in multiple jobs, and extensive tertax real interest rate effects) are reliance on incentive pay-pro­ Social Insurance, not sufficient to explain it. This er­ duces substantially higher levels of Incentives, and Risktaking ror may reflect a large downward productivity than do more "tradi­ Hans-Werner Sinn revision of permanent disposable tional" approaches involving nar­ NBER Wolking Paper No. 5335 income, which affected the con­ row job definitions, strict work November 1995 sumption of Swedish households rules, and hourly pay with close JELNo. HZ over and beyond the negative ef­ supervision. In contrast, adopting Public Economics fects of the drop in real asset pri­ individual work practice innova­ ces. We suggest that this revision tions in isolation has no effect on From the perspective of parents; in permanent income may have productivity. We interpret this evi­ redistributive taxation can be seen been triggered, at least in part, by dence as support for recent theo­ as social insurance for their chil­ the fiscal expansion of the early retical models that stress the im­ dren, for which no private alterna­ 1990s. portance of complementarities tive exists. Because private insur­ among a firm's work practices. ance comes too late during a per­ The Effects of Human son's life, it cannot cover the same Hiring Risky Workers risks as social insurance. Empirical­ Resource Management ly, 85 percent of social insurance Edward P. Lazear Practices on Productivity covers risks for which no private Casey Ichniowski, NBER Working Paper No. 5334 insurance would have been avail­ Kathryn Shaw, and November 1995 able. Redistributive taxation can Giovanna Prennushi JEL Nos. JOO, J6, J41 enhance effiCiency, because it cre­ NBER Wolking Paper No. 5333 Labor Studies ates safety and because it stimu­ November 1995 In finance and other literature it lates income-generating risktaking. JEL Nos. J4, J5, DZ has long been recognized that vari­ However, it also brings about detri­ Labor Studies, Productivity ance provides option value. The mental moral hazard effects. Both Increasingly, firms are consider­ same point extends to the labor the enhancement of risktaking and ing the adoption of new work market. Firms like variance in new the moral hazard effects tend to in­ practices, such as problem-solving employees because they can keep crease inequality in the economy teams, enhanced communication the good workers and terminate and, under constant returns to risk­ with workers, employment securi­ the bad ones. But market wages taking, this increase is likely to be ty, flexibility in job assignments, must adjust to make the marginal strong enough even to make the training workers for multiple jobs, firm indifferent between high- and net-of-tax income distribution more and greater reliance on incentive low-variance workers. unequal. Optimal redistributive tax­ pay. This paper provides empirical We explore the market equilibri­ ation either will imply that the pie answers to the question: do these um for new, risky workers to de­ becomes bigger when there is less human resource management prac­ termine how workers and firms inequality in pretax incomes or that tices improve worker productivity? line up on the various sides of the more redistribution creates more For this study,' we constructed market. Firms in growing industries posttax inequality. The welfare our own database through person­ prefer young, high-variance work­ state will encounter severe risks al site visits to 26 steel plants that ers. Growing industries are charac­ when free migration of people, used one specific steelmaking pro­ terized by high turnover rates. In goods, and factors of production cess. We collected longitudinal order for risky workers to provide becomes possible. Basing redistrib­ data with precise measures on pro­ option value, the initial employer utive taxation on a nationality prin­ ductivity, work practices, and the must have some advantage over ciple may help to overcome some technology in these production other firms. Private information or of these risks. lines. Our empirical results consis­ mobility costs can provide that ad­ tently support this conclusion: the vantage. It is also required that the adoption of a coherent system of risk have a firm-specific compo­ new work practices-including nent. General variations in ability

86. NBER RepotterFalll995 Procyclical Productivity: Historical Factors in A New Sample of Increasing Returns or Long-Run Growth Americans Linked from • Cyclical Utilization? the 1850 Public Use Micro Susanto Basu Sample of the Federal NBER Working Paper No. 5336 Census of Population to November 1995 Fertility and Marriage in the 1860 Federal Census JEL No. E2 New York State in the Era Manuscript Schedules Economic Fluctuations, of the Civil War Joseph P. Ferrie Monetary Economics Michael R. Haines and NBER Historical Paper No. 71 It has long been argued that AveryM. Guest August 1995 cyclical fluctuations in labor and NBER Historical Paper No. 70 JEL Nos. NOl, N31 capital utilization and the presence July 1995 Although the geographic, occu­ of overhead labor and capital are This paper analyzes a 5 percent pational, and financial mobility of important for explaining procycli­ systematic sample of households average Americans were important cal productivity. Here I present two from the man~cripts of the New aspects of economic development simple and direct tests of these hy­ York State Census of 1865 for sev­ in the nineteenth-century United potheses, and a way of measuring en counties (Allegany, Dutchess, States, the extent and correlates of the relative importance of these Montgomery, Rensselaer, Steuben, this economic mobility have re­ two explanations. The intuition be­ Tompkins, and Warren). The sam­ mained open to debate in the ab­ hind the paper is that materials in­ ple was selected to provide a vari­ sence of individual-level longitudi­ put is likely to be measured with ety of locations, settlement dates, nal data. This essay describes a less cyclical error than labor and and types of agricultural economy, new sample of 4837 individuals capital input, and materials are and includes two substantial urban linked from the 1850 Public Use likely to be used in strict propor- areas (the cities of Troy and Pough­ Micro Sample of the Federal Cen­ ~~ tion to value added. In that case, keepsie). This census was the first sus of Population to the 1860 Fed­ materials growth provides a good in the United States to ask a ques­ eral Census Manuscript Schedules, •_. measure of the unobserved chang­ tion about 'children ever born." using the new national 1860 Feder­ es in capital and labor input. Using These parity data, along with own­ al Census Index. The linked sam­ this measure, I find that the true children estimates of age-specific ple provides information on occu­ growth of variable labor and capi­ overall. and marital fertility rates, pation, wealth, family structure, tal inputs is, on average, almost are used to examine the relation­ and location in both 1850 and twice the measured change in the ship between fertility and rural-ur­ 1860. I describe in detail the COn­ capital stock or labor hours. More ban residence, occupation, ethnici­ Struction of the sample and test its than half of that is caused by the ty, literacy, and location within the representativeness, and I give ex­ presence of overhead inputs in state. Singulate mean ages at first amples of its potential uses. production; the rest is attributable marriage and other measures of to cyclical factor utilization. nuptiality also are estimated. The The Farm-Nonfarm Wage parity data provide direct evidence Gap in the Antebellum of fertility decline in the United United States: Evidence States during the first half of the from the 1850 and 1860 19th century. Township data are Censuses of Social added to the individual records to Statistics provide contextual variables. The Robert A. Margo' issue of ideational versus socioeco­ NBER Historical Paper No. 72 nomic and structural factors in fer­ August 1995 tility is discussed. JEL No. N31 Sectoral wage gaps for workers of comparable skill are central to issues in economic development • and economic history. This paper NBER RejJorterFalll995 87. presents new archival evidence on Fixing the Facts: Editing Percentiles of Modern the fann-nonfarm wage gap for the of the 1880 U.S. Census Height Standards for Use United States just prior to the of Occupations in Historical Research American Civil War. Measured at with Implications for Richard H. Steckel the level of local labor markets, the Long-1erm Trends and NBER Historical Paper No. 75 wage gaps are small and not very the Sociology of October 1995 persistent over time. Aggregated to Official Statistics reflect the geographic distribution Susan B. carter and Percentiles of modern height of farm and nonfarm labor, the Richard Sutch standards are useful in historical re­ search because children differ sys­ gaps are larger than previously NBER Historical Paper No. 74 tematically in height by age, and thought. I also show that invest­ October 1995 ment in manufacturing capital be­ there are differences in growth po­ tween 1850 and 1860 responded to We argue that the enumerators' tential by gender and perhaps labor market inefficiencies indicat­ occupational returns from the im­ across some ethnic groups. Modem ed by the gaps: counties with rela­ portant Census of 1880 were edited height standards are needed to tively low fann wages experienced heavily prior to publication. The make relative comparisons of nutri­ above-average investment. effect was to substantially reduce tional status in these circumstances. the number of individuals reported The standards also are used to as­ Myth of the Industrial with an occupation. For youthful· sess progress or deprivation against Scrap Heap: A Revisionist and older males and all women, a level that we know is attainable View of Turn-of-the­ the editing was so substantial as to under good environmental circum­ Century American qualitatively affect the apparent stances. Historical researchers who Retirement trend in labor force participation need modem height standards en­ Susan B. carter and for these groups over time. The counter several problems, includ­ Richard Sutch stylized facts regarding labor mar­ ing the choice of standards, manip­ NBER Historical Paper No. 73 ket dynamics during the period of ulation of those standards to meet American industrialization and the the requirements of historical data, October 1995 historical stories constructed around and calculation of percentiles. After Using the census survival meth­ them now will need to be reexam­ discussing criteria used in selecting od to calculate net flows across ined. We contend that the editing standards, which lead to the choice employment states between 1900 was authorized secretly by Francis of NCBS heights as a reference, and 1910, we find that approxi­ Amasa Walker, superintendent of this paper gives percentiles calcu­ mately one-fifth of all men who the Tenth Census of 1880 and one lated in line with the requirements reached the age of 55 eventually of the most prominent and deco­ of historical data. Results are stated retired before their death. Many of rated economists, statisticians, and in centimeters and inches, and by these retirees appear to have public servants in America at that age at last birthday and age at planned their withdrawal from paid time. While other scholars have nearest birthday. employment by accumulating as­ identified potential sources of bias sets, becoming self-employed, and in Census figures, no one has here­ then liquidating their assets to pro­ tofore suggested that the official vide a stream of income to finance statistics of the United States were consumption in old age. This altered· covertly to present a picture "modem" retirement behavior, we different from information collected argue, has important implications by Census enumerators. If we are for the economic history of capital correct, the socioiogy of official and labor markets, saving and in­ nineteenth-century American statis­ vestment, insurance and pensions, tics will require rethinking. and the family economy.

88. NBER Repon-er Fall 1995 • Technical Papers Labor Studies, Productivity of survivors, even if the original assignment is random. Similarly, We develop a model for decom­ randomized training interventions, posing the covariance structure of such as National Supported Work, One Day in June 1993: panel data on firms into one part are not necessarily assigned ran­ A Study of the Working of caused by permanent heterogene­ domly in the sample of working Reuters 2000-2 Electronic ity, one part caused by differential men. A nonexperimental version of Foreign Exchange histories with unknown ages, and this problem involves the use of Trading System one part caused by the evolution instrumental variables (IV) to esti­ Charles Goodhart, of economic shocks to the firm. mate behavioral relationships. A Takatoshi Ito, and Our model allows for the endoge­ Richard Payne sample selection rule that is related nous death of frrms, and correctly to the instruments can induce cor­ NBER Technical Paper No. 179 handles the problems arising from April 1995 relation between the instruments the estimation of this death pro­ and unobserved outcomes, possi­ ]EL Nos. F3, G2 cess. We implement this model on Asset Pricing, International Finance bly invalidating the use of conven­ an unbalanced"longitudinal sample and Macroeconomics tional IV techniques in the selected of French firIns that have both sample. This paper shows that con­ This paper uses foreign ex­ known and unknown ages and his­ ditioning on the probability of change data (bid, ask, and transac­ tories. For firms with unknown selection given the instruments can tion prices and quantities) collected birthdates, we find, the structural provide a solution to the selection from the electronic broking system, autocorrelation in employment, problem as long as the relationship Reuters D2000-2, on June 16, 1993. compensation, and capital is domi­ between instruments and selection We compare the bid and ask nated by the part caused by initial status satisfies a simple monotonic­ quotes, which are "firm" in this da­ heterogeneity and random growth ity condition. A latent index struc­ taset, with the Reuters FXFX page, rates. Serial correlation in the peri­ ture is not required for this result, which reports only indicative bid odic shocks is less important. For which is motivated as an extension and ask prices. The samples we these firms, profitability, value add­ of earlier work on the propensity have are small: only seven hours. ed, and indebtedness have process­ score. I illustrate the conditioning Still, we find that although the bid­ • es in which the heterogeneity com­ approach to selection problems ask mean of indicative quotes is ponents are less important. Firms using IV techniques to estimate the similar to that of "frrm" quotes, the with known birthdates and histo­ returns to schooling in a sample behavior of the bid-ask spread and ries (which are younger than the with positive earnings. the frequency of quote entry are censored firms) have autocorrela­ quite different in the two types of tion structures dominated by the Refining Estimates of quotes. The bid-ask spreads in the heterogeneity. broking system vary much more Marital Status Differences in Mortality at Older Ages with time and depend on the fre­ Conditioning on the quency of trade, while the indica­ Sanders Korenman, Probability of Selection Noreen Goldman, and tive bid,.-ask spreads tend to cluster to Control Selection Bias HaishanFu at round numbers. Joshua D. Angrist NBER Technical Paper No. 182 NBER Technical Paper No. 181 July 1995 A la Recherche des June 1995 Moments Perdus: ]ELNos.112,]12,]14 JEL Nos. C24, ]31 Aging Covariance Models for Labor Studies unbalanced Panels with The main objective of this analy­ Endogenous Death Problems of sample selection sis is to demonstrate that some of John M. Abowd, Bruno Crepon, arise in the analysis of both experi­ the limitations that have character­ Francis Kramarz, and mental and nonexperimental data. ized recent studies of the relation­ Alain Trognon In clinical trials to, evaluate the im­ ship between marital status and NBER Technical Paper No. 180 pact of an intervention on health health outcomes may result in bi­ May 1995 and mortality, treatment assignment ased estimates of marital status dif­ • ]EL Nos. C33,]0 is typically nonrandom in a sample ferences in mortality among the el-

NBER Reporter Fall 1995 89. derly. Our secondary goal is to Randomization as an treatment is the same for everyone evaluate the evidence supporting Instrumental Variable (conditional on observables), or the the excess risks of mortality associ­ James J. Heckman treatment effects are variable but ated with widowhood, once we NBER Technical Paper No. 184 the person-specific component of are able to eliminate or mitigate September 1995 the variability that cannot be fore­ many of the limitations that affect­ JEL Nos. C13, C93 cast by observables does not deter­ ed other studies. Labor Studies mine participation in the program, Our results, based on the 1984- widely used instrumental variable 90 Longitudinal Study of Aging, This paper discusses how ran­ methods do not produce consistent demonstrate that the estimated ef­ domized social experiments oper­ estimators of the parameter of in­ fects of marital status in logit and ate as an instrumental variable. For terest. Neither assumption is very hazard models of survival are very two types of randomization schemes, palatable. The first assumes a ho­ sensitive to whether and how in­ the fundamental experimental esti­ mogeneity that is implausible. The formation on marital status is up­ mation equations are derived from second assumes either very rich dated after the baseline interview. the principle that experiments data available to the econometri­ Refined measures of marital status equate bias in control and experi­ cian, or that the persons being that prospectively capture transi­ mental samples. Using convention­ studied either do not have better tions from being married to wid­ al econometric representations, I information than the econometri­ owhood result in substantially larg­ derive the orthogonality conditions cian or do not use it. Instrumental er estimates of the relative risk of for the fundamental estimation variable methods do not provide a dying in the early stages of widow­ equations. Randoinization is a mul­ general solution to the evaluation hood (bereavement). tiple instrumental variable, in the problem. . sense that one randomization de­ Another Heteroskedasticity fmes the parameter of interest ex­ Information-Theoretic and Autocorrelation pressed as a function of multiple Approaches to Inference Consistent Covariance endogenous variables in the con­ in Moment Condition ..... Matrix Estimator ventional usage of that term. It or­ Models 'WI Kenneth D. West thogonalizes the treatment variable Guido w. Imbens, NBER Technical Paper No. 183 simultaneously with respect to the Phillip Johnson, and July 1995 other regressors in the model and Richard H. Spady Asset Pricing the disturbance term for the condi­ NBER Technical Paper No. 186 I propose and evaluate a --IT tional population. However, con­ October 1995 consistent estimator of a het­ ventional "structural" parameters in Labor Studies eroskedasticity and autocorrelation general are not identified by the The recent papers of Back and consistent covariance matrix esti­ two types of randomization schemes Brown (990), Imbens (993), and mator. In the relevant applications used widely in practice .. Qin and Lawless (994) develop the regression disturbance follows a moving average process of Instrumental Variables: one-step efficient GMM estimation. known order. In a system of t A Cautionary Tale These papers emphasize methods equations, this "MA-t" estimator en­ James J. Heckman that correspond to Owen's (988) empirical likelihood to reweight tails estimation of the. moving av­ NBER Technical Paper No. 185 the data so that the reweighted erage coefficients of an t-dimen­ September 1995 sample obeys all the moment re­ sional vector. Simulations indicate JEL Nos. C14, C31, C35 strictions at the parameter esti­ that the MA-t estimator's finite sam­ Labor Studies ple performance is better than that mates. In this paper, we consider of the estimators of Andrews and This paper considers the use of an alternative KLIC-motivated Monahan (992) and Newey and instrumental variables to estimate weighting and show how it and West (994) when cross-products the mean effect of treatment on the similar discrete reweightings defme of instruments and disturbances are treated. It reviews previous work a class of unconstrained optimiza­ sharply negatively auto correlated; on this topic by Heckman and tion problems, which includes otherwise they are comparable or Robb 0985, 1986) and demon­ GMM as a special case. Such KLIC- .. slightly worse. strates that unless the effect of motivated reweightings introduce ,.

90. NBER Reporter Fall 1995 ·e,. M auxiliary "tilting" parameters, A CES Indirect where M is the number of mo­ Production Function ments; parameter and overidentifi­ Boyan Jovanovic cation hypotheses can be recast in NBER Technical Paper No. 188 terms of these tilting parameters. October 1995 Such tests, when conditioned ap­ JELNo.031 propriately on the estimates of the Productivity original parameters, are often start­ lingly more effective than their This paper derives an indirect conventional counterparts. This is production function that is, in a apparently because of the local an­ special case, of a constant elasticity cillarity of the original parameters of substitution (CES) form. This is for the tilting parameters. not a contribution to the theory of aggregation generally. Instead it is Selection Bias Adjustment a microfoundation for a specific but in Treatment-Effect Models popular production function-the as a Method of CES-that help~ us express the im­ Aggregation portant concept of the elasticity of Robert A. Moffitt substitution in terms of more primi­ NBER Technical Paper No. 187 tive, and more intuitive concepts of October 1995 the returns to scale. The paper pre­ JEL No. C3 sents a simple lemma, and then Public Economics shows that several and diverse ap­ plications have a common logical This paper aims to interpret es­ structure: the production function timation of the conventional treat­ often used in growth theory, the ment-effect selection-bias model in utility function when there is t:. econometrics as a method of aggre­ household production, human cap­ gation, and to draw the implica­ ital theory, and the concept of the tions of this interpretation. In addi­ aggregate technology shock. tion, the paper notes the connec­ tion between this interpretation and I an older style of analysis using grouped data, and illustrates the aggregation analogy with examples from the literature. The method of I instrumental variables is the estima­ tion technique used to illustrate the points.

NBER Reporter Fall 1995 91. Nonprofit Org. • U.S. Postage Paid Boston, MA Permit No. 55932

1050 Massachusetts Avenue Cambridge, Massachusetts 02138-5398 (617) 868-3900

Address Correction Requested

:0;"Ff,J,',,,;S,,sY,/c0!biill&,2k\;Q2,i,{,,;j,,,,,)},,,,),);>, >, "'>i::.:z2&::.:z:B!.l@ill@illili1~2&i.JE~ili1~iJEj~~[iill~~~~~~~~~~~':il!,\if;:" "" ,